# Kamu, full text

> Kamu is a growth agent for SaaS founders, doing the selling and the marketing both. You add your product by its URL, and Kamu reads your site and your competitors, works out who is likely to buy, and then does the work every day on its own: finding people who fit, writing them cold email and LinkedIn messages, answering the replies, posting, drafting SEO articles, filing directory listings, building its own landing page, and making video ads it leaves paused for you to start. Every action it takes is logged with the reason behind it.

Canonical site: https://kamugrows.com
Link map: https://kamugrows.com/llms.txt

# What Kamu is

Kamu is a paid SaaS product, not an agency and not a chat assistant you have to prompt. It runs on its own schedule without anyone starting it, and it keeps working while the founder builds.

It is built for solo founders and small SaaS teams who would rather ship product than do marketing. The thing it replaces is a growth hire, not a marketing tool: it decides what to do, does it, and reports back, rather than waiting to be told.

What makes that safe is that the limits are numbers rather than promises. Each plan has a hard daily cap per channel, the founder chooses an overall pace, and the plan carries the credits Kamu spends doing the work. Those are enforced in code, so Kamu stops at them whether or not the model would have chosen to. Anything that spends money or changes the plan arrives as a decision to approve or decline.

The founder remains the sender of everything Kamu sends and the controller of the prospect data it processes. Kamu operates the accounts; it does not own them.

# How it works

1. Add your product by URL. Kamu reads the site, works out who buys something like it, and looks at the companies you are up against.
2. Kamu writes a growth strategy in plain words and keeps every version, so you can read what it is working on and what changed.
3. It works the plan daily, inside the plan's hard caps.
4. You get a standup each morning, a decision inbox for the calls worth your judgment, and an activity log for everything else.

# Channels

Connecting a channel is optional. When one is not connected Kamu still writes for it: drafts land in the queue for you to publish yourself, and Kamu switches to publishing on its own the moment you connect the account.

- Cold email: sends from Kamu's own pre-warmed domain by default, or from a subdomain of your own once you verify it. Unsubscribe handling, suppression lists, daily caps, domain warmup, and a bounce and complaint circuit breaker are enforced in code.
- SEO content: always on. Drafts land in a content library, and Kamu can publish to your CMS once you connect one and approve the draft.
- X: Kamu posts through X's own API once you connect the account, and replies to people who mention you. X allows automated replies nowhere else, so those are drafted for you.
- LinkedIn: posts and direct messages, through one hosted account connection.
- GitHub: read-only. Kamu reads releases and merged pull requests from the repositories you share, so posts and articles about recent work are built on what actually shipped. It opens no issues, files no pull requests and pushes nothing.
- Instagram: posts to your feed through that same hosted connection, each with a picture Kamu makes in your colours, because Instagram will not take a caption on its own. There is no official posting API for this and Kamu says so plainly before you connect. Each connected account counts against your plan's account cap.
- Reddit: draft-only, always. Reddit's commercial API is priced for enterprises, so Kamu reads threads publicly, writes the comment, and you post it.
- Slack: chat with Kamu, approve decisions, and get the daily standup in a channel.

# Pricing

Four plans, billed monthly in euros. Seed is the way in: twenty-five euros a month, and the first month grants double the credits. Each includes credits, which is what Kamu spends doing the work: roughly one cent each, so about 2 credits to find a lead, 25 for an article, 100 for a video ad and nearer 200 for the first few, while Kamu builds its cast of presenters. Run out and Kamu stops and says so rather than running up a bill. More can be bought at any point and bought credits never expire, while the ones included with a plan reset when it renews. Pay yearly and two months are free. Every plan includes the activity log, the decision inbox, and all six objectives.

### Seed, €25 a month

- 1 product
- 1,000 credits a month
- 10 outreach emails a day
- 2 LinkedIn DMs a day
- 2 X posts a day
- 1 Reddit actions a day
- 1 content drafts a week

### Starter, €49 a month

- 1 product
- 2,500 credits a month
- 25 outreach emails a day
- 5 LinkedIn DMs a day
- 5 X posts a day
- 3 Reddit actions a day
- 5 content drafts a week

### Growth, €149 a month

- 3 products
- 7,500 credits a month
- 75 outreach emails a day
- 15 LinkedIn DMs a day
- 10 X posts a day
- 5 Reddit actions a day
- 10 content drafts a week

### Scale, €349 a month

- 10 products
- 17,500 credits a month
- 200 outreach emails a day
- 25 LinkedIn DMs a day
- 25 X posts a day
- 10 Reddit actions a day
- 30 content drafts a week

## Questions about pricing

Q: What is a credit?
A: One credit is roughly one cent of what Kamu spends growing your product. Finding a lead is about 2 credits and an article about 25. Video costs depend on the script and rendering workflow; Kamu checks the quote against your remaining credits before starting.

Q: Can Kamu grow a mobile app?
A: Yes. Add an App Store or Google Play link and choose Mobile app. Kamu researches the listing and audience, then works on content, social posts, UGC videos and app discovery. Store listing changes stay as recommendations for you to apply.

Q: What happens when I run out?
A: Kamu stops and tells you, in its morning note and in the app. Nothing is charged that you did not already pay for, and nothing keeps running up a bill in the background. You can buy more credits at any point and it picks straight back up.

Q: Do unused credits carry over?
A: The credits included with your plan reset when it renews. Credits you buy on top never expire, and Kamu spends the monthly ones first so the ones you paid extra for are the last to go.

Q: Do the prices include VAT?
A: No, they are before VAT. It is worked out at checkout from where you are and added there, so the total you approve is the total you pay. If your business is VAT registered in the EU outside Finland, the reverse charge normally applies and you account for the VAT in your own return instead of paying it to us.

Q: Can I change plans later?
A: Yes. Move up or down whenever you like, and the change takes effect on your next cycle.

Q: Is there a free trial?
A: No. You pick a plan before Kamu starts working, and you can cancel whenever you like.

# What Kamu does, in full

## Outreach

Page: https://kamugrows.com/features/outreach

I write your cold email one at a time. I find people who match your buyers, read what they said, and check the address before I write. When they reply, I answer. When they say no, I stop.

### Where I get the names

I read your product, work out who buys something like it, and go looking for those people. Where I can read what someone posted, I do that first, so the opener points at something real.

- People who match the buyers in your plan
- Every address checked before anything goes out
- One row per person, however many addresses they have

### When someone writes back

I pick the reply up on my next run, read the whole thread, and answer the question they asked. Leave a note on a thread and I take it from there.

- Follow-ups sit days apart, with a cap per person
- Your meeting link goes in when someone asks for it

### I can tell a no from a never

"We already use something" closes the thread. "Take me off your list" suppresses the address across every product you run. Neither gets another email from me.

- A no closes the thread
- An unsubscribe is honoured across every product you run
- I record whether each email was delivered

### Email you would put your name on

- People who match your buyers: I learn who says yes, and go looking for more of them.
- Checked addresses: I check the address works first, so you don't collect bounces.
- Short and specific: Each one points at something that person shipped or wrote.
- I answer the reply: I read the whole thread and write back on my next run.
- Paced like a person: Follow-ups sit days apart, and stop at the cap you set.
- One place per person: Everything you and I sent, and everything they said back.

### How outreach works

1. I learn your buyer: I read your site and your plan to work out who is worth writing to.
2. I find them and check: New people who fit, each with an address that actually works.
3. I write, then keep writing: A first email worth answering, then the rest of the conversation.

### Questions about outreach

Q: Will it email people without my say?
A: Your plan has a hard daily cap, and I stay inside it. Pause me whenever you want, or leave a note on a thread to change how it goes.

Q: What happens when someone replies?
A: I read the full thread and answer in context. If the timing is wrong I say so and come back later. Every message is in the same conversation.

Q: What if someone asks to be left alone?
A: I stop. An unsubscribe is honoured across every product in your account. A plain "not interested" closes that conversation without suppressing the person.

Q: How do I know the emails are arriving?
A: I record delivery for every send, so "nobody replied" and "it never landed" are different lines. There is no open tracking, because a pixel is what makes a plain email look like marketing.

## SEO content

Page: https://kamugrows.com/features/content

I write you an article a week, aimed at a real search. I find the phrases your buyers type into Google, pick one we can rank for, and write the whole draft. You read it and press publish.

### I pick the search term before I write

I look at what people search for around your product, how hard those pages are to beat, and whether anyone typing the phrase would buy. Each draft targets one phrase, in the shape that kind of page wants.

- Search terms from your own market
- One target phrase per article
- Headings, links and a conclusion, ready to read

### You keep the publish button

Every piece lands as a draft. Connect Ghost or WordPress and one button puts it live on your site. Without a CMS you copy the markdown out.

- A draft a week, or slower if you would rather
- Every keyword carries the reasoning that chose it

### What you get without hiring a writer

- Terms with buyers behind them: I write for phrases people type when they want to buy something.
- Finished drafts: Headings, links and a conclusion. Not an outline for you to fill in.
- No AI tells: Every draft is run past about 120 banned phrases and rewritten where it trips.
- Publishes to your CMS: Approve a draft and one button puts it live on Ghost or WordPress.
- A pipeline you can read: Each phrase carries its intent, its difficulty, and why I picked it.
- One a week: A pace you can keep up with while you build the product.

### How content works

1. I find the phrase: I look at what your buyers search, and what I can realistically beat.
2. I write the draft: One target phrase, in the shape that kind of page wants.
3. You publish: Approve it and I can put it live, then watch how it does.

### Questions about seo content

Q: Do I have to publish what you write?
A: No. Every piece lands as a draft. Approve it and I publish to your CMS, or copy the markdown out and put it wherever you like.

Q: Will it read like generic AI content?
A: I read your product first, and every draft is checked against about 120 banned phrases and a set of shape rules before you see it. Some drafts still want a pass from you.

Q: Which search terms do you target?
A: Ones your buyers type when they are shopping, weighed against how hard the page is to beat. I skip the big vague terms I have no chance at.

Q: Can you publish automatically?
A: With Ghost or WordPress connected, yes, after you approve the draft. Nothing goes live while it is still waiting for your approval.

## Social

Page: https://kamugrows.com/features/social

I post about what you actually shipped. I write for X, LinkedIn, Instagram and Reddit the way a founder does: one thing that happened this week, in plain words. Connected accounts get the post. The rest get a draft.

### I write about the week you actually had

Connect GitHub and I read what shipped off your releases and merged pull requests. Each platform gets its own version, and I compare every draft against your last few posts so they stop sounding alike.

- Written separately for each platform
- Built from your own releases and merges
- Checked against a list of AI tells before I post

### I respect the room

I read a subreddit's rules before I write and only touch communities you approved. Reddit's API is priced for enterprises, so there I write the comment and you post it. If nothing is worth posting, I don't post.

- Only the accounts and subreddits you approve
- A hard daily cap for each channel

### What showing up every day looks like

- Written per platform: An X post and a LinkedIn post are different things, so I write both.
- No AI tells: About 120 banned phrases, checked and rewritten before anything goes out.
- I read the rules: Every subreddit has its own. I check before I write.
- Hard daily caps: Every plan has a clear ceiling. Pause social whenever you want.
- About real things: Read off your own repository: what shipped, not what I imagined shipped.
- All of it logged: Every post, with the reason I wrote that one.

### How social works

1. You connect accounts: Link X, LinkedIn or Instagram, and name the subreddits you want to be in.
2. I write: One post per platform, from what happened this week.
3. I post, or hand you the draft: X, LinkedIn and Instagram go out on their own. Reddit comes to you to post.

### Questions about social

Q: Will you spam subreddits?
A: No. I only write for subreddits you approve, read their rules first, and stay under the hard daily cap. I answer the question before I mention you.

Q: How do you know what I shipped?
A: Connect GitHub and I read your releases, or merged pull requests when a repository has none. I never write anything back. On a week where nothing shipped I say so.

Q: Which platforms are supported?
A: X, LinkedIn, Instagram and Reddit. I post to the first three once you connect them. Reddit's API is priced for enterprises, so there I write the comment and you post it.

Q: How does Instagram work?
A: You connect once through a hosted window and I post to your feed with a picture in your colours. There is no official API for a personal account, so I drive yours the way a phone app would, with low daily caps.

## Video ads

Page: https://kamugrows.com/features/ads

I build the video ad and leave it paused. I write the script, cast a presenter, render a short vertical video, and build the Meta campaign around it. Everything stays paused until you start it.

### Nobody in the video claims to use your product

The person on screen is a presenter, never a customer. They can say what the product does. They cannot claim to use it or to have got a result from it, and a script that tries is refused before anything renders.

- A cast of six presenters, taking turns
- Your product named once, plainly
- No invented numbers, no fabricated testimonials

### The campaign arrives built and paused

Campaign, ad set and ad, all paused, with the copy and targeting written around the video. Without a connected ad account I draft the whole thing for you to run by hand. Nothing I create can start spending.

- Every campaign, ad set and ad is created paused
- A ceiling on the daily budget, checked in code
- I can pause a campaign, but never start one

### I wait before I judge it

Once it runs I read back what it spent, how many people saw it, how many clicked, and what a click cost. No verdict under a thousand impressions. An ad that spent real money and got no clicks is paused without asking.

- No verdict under a thousand impressions
- Spend, clicks and cost per click, in the morning standup

### What I will and will not do with your ad money

- The video itself: Script, presenter, voice and captions, from what your product does.
- Always paused first: I build the campaign. Only you can start it.
- A ceiling in the code: A hard cap on the daily budget, checked before the request goes out.
- A cast of six: Presenters and rooms take turns, so the ads do not blur together.
- Judged on clicks: What it spent and what a click cost, once a thousand people have seen it.
- Watch before you spend: Every video plays and downloads from your library, connected to Meta or not.

### How video ads work

1. I make the video: I write the script, cast a presenter and render it with captions.
2. I build the campaign: Targeting, copy and a daily budget around that video, all paused.
3. You press go: In Ads Manager, when you are ready. I read the numbers from there.

### Questions about video ads

Q: Can you spend my ad money on your own?
A: No. I cannot build a campaign until you approve the decision to run one, and everything I create is paused, so only you can start it. There is also a hard ceiling on the daily budget.

Q: Who is the person in the video?
A: A generated presenter, not a customer. A cast of six take turns. They can say what the product does, but not that they use it, and a script that tries is refused.

Q: Do I need a Meta account connected?
A: No. Without one I still make the video and draft the whole ad for you to run by hand. Connect an ad account and I build it in there, paused.

Q: What does a video cost?
A: Agent Media quotes each clip before I submit it. I check that quote against your remaining credits and daily budget. Ad spend is separate and sits on your own Meta account.

## Strategy

Page: https://kamugrows.com/features/strategy

I write the plan down before I start. Before I send anything, I read your site and the companies you are up against, and write the plan down. You get a few sentences. The long version is a click away.

### I do the homework first

I read your product the way a new head of growth would on their first day: what you do, who it is for, who else sells to those people. The plan comes in two lengths. The short one is for you. The long one is what I work from.

- I read your site, then the companies selling near you
- A few sentences for you, the long version underneath

### The plan changes when the numbers do

I revisit it as replies and posts come back, and move weight toward whatever is working. When a call is big enough to be yours, I write it up and ask.

- Every version of the plan is kept
- Big calls come to you as a question

### Why I do what I do

- I read first: Your product, your buyers and your competitors, before anything else.
- A plan you can read: Plain words in a document you can open any time.
- Everything traces back: Each action points at a line in the plan.
- Old versions stay: You can see what I used to think, and when I changed my mind.
- I ask about the big ones: Spend, positioning, a new channel. Those come to you first.
- A reason on every action: Why I wrote that email, tied back to the plan.

### How strategy works

1. I read: Your site, your market, and the companies selling near you.
2. I write the plan: A wedge, the people to go after, and the channels to use.
3. I work and adjust: Every day, and I move weight toward whatever replies.

### Questions about strategy

Q: Can I see the plan?
A: Yes. I write it in plain words and keep every version, so you can always read what I am working on, what changed, and which result caused the change.

Q: What if I disagree with your approach?
A: Leave me a note and I take it from there. Big calls also come to you as a decision to approve or decline, and a declined proposal is dropped.

Q: How does the plan change?
A: As replies and posts come back, I revisit it and shift weight toward whatever is working. I write down what changed and why.

Q: Are you just guessing?
A: I read your product and your market first, and every action carries the reason behind it. Early on I am working from less, and I say so.

## Results

Page: https://kamugrows.com/features/results

See everything I did today. Every email, post and draft is logged with the reason behind it, next to the numbers that say whether any of it is working.

### Nothing happens off the record

I write down every action as I take it, with the reason that led there. A whole day takes about a minute to read, so you never have to guess what went out under your name.

- Every action carries the reason behind it
- Filter by channel when you only care about one

### Four numbers, and what they cost

How many people wrote back, and how many of those look like buyers. Under those, what got published and what the month cost. Run out of credits and I stop and tell you.

- Replies and qualified leads, not impressions
- Every credit spent, and what it was spent on

### Everything is on the record

- The full log: Every action I take, with the reason behind it.
- Replies and signups: How many people wrote back, and how many look like buyers.
- No surprise bill: I spend the credits your plan includes, then stop and ask.
- Filter by channel: Read only the email, or only the posts.
- The why, attached: Every action says why I chose to do that one.
- Updates as I work: The feed moves while I am running.

### How results work

1. I work: Email, social and content, on my own schedule through the day.
2. I log everything: Each action lands on the timeline with the reason for it.
3. You read it: The work and the numbers, whenever you feel like checking.

### Questions about results

Q: How do I know what you're doing?
A: Everything I do lands on the activity timeline, each action with the reason behind it. Filter by channel, or open an item to inspect the work itself.

Q: Can I see the cost?
A: Yes. Every credit I spend is on the ledger with what it went on, and the balance moves on your billing page. When the included credits run out, I stop.

Q: Which numbers do you track?
A: Emails sent, replies, people who look like buyers, content published, and the credits spent getting there. The result stays next to the work that produced it.

Q: Is the feed live?
A: It updates while I am running, so you can check in and see what happened ten minutes ago. Completed work, drafts and decisions all stay in the same timeline.

## Works like a hire

Page: https://kamugrows.com/features/colleague

I work like someone you hired. I learn how you write and write that way. I start the day with a plan, work it, and bring you the calls worth your judgment. Each morning you get a short note from me.

### I write like you, not like a tool

I read what you have already written, your posts, about page and changelog, and measure what can be counted: sentence length, lowercase openers, whether you have ever used a hashtag. A draft that does not sound like you is rewritten, not sent.

- Learned from your own writing, not from a description of it
- I read your site for how the company talks about itself
- A draft that does not sound like you gets rewritten

### I tell you what I'm up to

Each morning, a short standup in my own voice: what I am on today, what I finished yesterday, anything that went wrong. On a quiet day I say the day was quiet.

- A standup in plain words every morning
- In Slack, Telegram or WhatsApp, or in the app

### I ask before I spend your money

A sponsorship, a change in positioning, a new channel. I write the call up with my reasoning and put it in your inbox. Approve it and I run with it. Decline it and I drop the idea for good.

- Approve or decline in one tap
- The reasoning arrives with the question

### How much I do without you

- I sound like you: Learned from your own posts and pages, down to whether you use hashtags.
- A standup every morning: A short note on the day's plan, in my own words.
- Decision inbox: The calls worth your judgment, written up and waiting.
- Approve or decline: I respect both, and remember what you turned down.
- Reasoning attached: Every proposal says why I think it is worth doing.
- I run while you sleep: What happened at 3am is in the morning note.

### How I work with you

1. I plan the day: A short standup says what I am working on.
2. I do the work: Email, social and content, on my own.
3. I bring you the big calls: Anything that spends money or changes the plan.

### Questions about works like a hire

Q: Will you do things without asking?
A: The everyday work, yes. That is the job. Anything that spends money or changes the plan comes to you as a decision first, and the tool itself enforces that boundary.

Q: What's in the daily standup?
A: A short note in plain words: what I am working on today, what I finished yesterday, and anything that went wrong. It can arrive in Slack, Telegram or WhatsApp as well as the app.

Q: What happens when I decline a proposal?
A: I drop it and do not bring it back. Approve one and I get on with it. Either choice is recorded, so the next run does not ask the same question again.

Q: How do you learn my voice?
A: I read writing that is already yours and measure what can be counted: sentence length, lowercase openers, hashtags, emoji. What I decided is written down on your strategy page, so you can correct me.

# Data and sub-processors

Kamu processes prospect and outreach data on the customer's behalf: the customer is the controller, Kamu is the processor. The third parties involved are:

- Agent Media: AI video generation from scripts and presenter images. Provider-managed processing regions; not specified in the API documentation.
- Anthropic: LLM powering the agent and in-app chat. United States.
- Resend: Outbound & inbound email, sending-domain DNS. United States.
- Hunter.io: Email finding & verification. European Union (France).
- Exa: Web search for prospect discovery. United States.
- Stripe: Subscription billing & payments. United States.
- Railway: Application, worker and managed PostgreSQL hosting. United States.

The full list, and the terms behind it, are at https://kamugrows.com/subprocessors and https://kamugrows.com/dpa.

# Blog

## A SaaS marketing strategy for a small team

https://kamugrows.com/blog/a-saas-marketing-strategy-for-a-small-team
Published 2026-09-10.

A SaaS marketing strategy should fit the people and time you have. Use replies to learn the market, then add search, directories and paid tests.

A SaaS marketing strategy for a small team has one job: create enough contact with possible buyers to learn why they act, then turn what you learn into work that can keep running.

That rules out the usual channel buffet. A founder with four hours a week cannot publish every day, run ads, write search articles, host a webinar and keep five social accounts alive. Trying produces a week of activity followed by a month of silence. Pick one channel that starts conversations now and one asset that compounds later.

For most young SaaS products, that means direct outreach beside search content. Outreach produces replies before the positioning is settled. Search turns the language in those replies into pages that can keep bringing the right person back.

## Start with the constraint you actually have

Budget matters, though attention usually runs out first. Write down how many hours someone can give marketing each week and who owns them. If the answer is "whoever has time," the work has no owner.

Then choose a buyer narrow enough to recognise. "Small businesses" is not a buyer. "The person running support at a Shopify store with five to twenty staff" is close enough to find, write to and learn from. The point is not to prove that this segment will be right forever. It is to make the first month capable of teaching you something.

Give that buyer one reason to care now. A recent hire, a product launch, a job posting, a new market or a public complaint can supply the timing. Company size and job title tell you who someone is; a [buying signal](/glossary/buying-signal) tells you why this week might be different from last month.

## Use the first month to learn the words

The early goal is a qualified conversation, since a reply carries more information than a page view. Write to a small number of people who fit the same buyer definition, keep the message plain, and note where they disagree.

Replies expose the parts of the pitch that a landing page hides. Someone may understand the problem but describe it with a different phrase. They may already solve it by hand and dislike the cost rather than the effort. They may care about a risk you treated as a footnote. Those are edits to the strategy, not objections to route around.

Keep one document with the phrases people use, the events that made them answer and the reasons they said no. After twenty real conversations, that document is worth more than a generic content calendar because it contains the market's language.

[Founder-led sales](/glossary/founder-led-sales) is useful here even if the founder never wants a sales job. The founder can change the product, price or audience after one call. A hired operator has to bring the same lesson back through a meeting and wait for a decision.

## Build the search layer from those conversations

Search starts after the message has met a few buyers. Take the questions that appeared in replies and sort them by intent.

A definition belongs in a glossary page. A product comparison needs an honest account of where each option wins. A practical question deserves a guide that lets the reader do the job, even if they never buy anything. Keep those page types separate so two URLs do not fight for the same query.

The first paragraph should answer the search. Do not make a reader cross an origin story to reach the thing they asked for. Add evidence where the claim needs it, show the date on facts that change, and link each article to the product page or reference page it supports.

One strong page a month is enough to start. The useful test is whether you would send it to a buyer who asked the question on a call. If sending it would feel like handing them a padded brochure, it should not be published.

## Put directories beside search

Directory listings are small jobs with two possible returns: a buyer may find the product there, and the listing may give a search engine another place to confirm what the company is.

Choose directories that serve the buyer or the product category. Fill out the profile fully, keep the description consistent with the site and record the live URL. A list of two hundred submission sites is mostly a list of forms. The useful set is much smaller.

This work pairs well with the search layer because both need the same facts: category, customer, short description, screenshots and a stable product URL. Do the facts once, then adapt them to each directory's format.

## Add paid tests after the page can carry them

Paid acquisition buys speed. It also sends more people through whatever is already confusing.

Wait until the landing page has a clear buyer, a specific promise and a next step you can measure. You should know what counts as a qualified trial and what share of those trials stay long enough to matter. Otherwise the ad platform will optimise for the easiest action, which is often the least useful one.

Start with a capped test around one message and one audience. Keep the campaign paused until a person has checked the creative, destination and budget. If the page does not convert, change the message or the page before raising spend.

## Measure the handoff between stages

A dashboard can make weak marketing look busy. Use a short chain of measures that connects the first contact to a customer:

| Stage | What to count | What it tells you |
| --- | --- | --- |
| Outreach | Qualified replies | The buyer and timing were close enough to start a conversation |
| Website | Trials or enquiries from the target segment | The page made the product clear enough to try |
| Product | Accounts still active after the first useful period | The channel brought people the product could keep |
| Spend | [Cost per qualified lead](/glossary/cost-per-qualified-lead) | The cost of reaching someone worth talking to |

Follower counts and impressions can diagnose distribution, but they do not complete this chain. A small audience containing buyers is more useful than a large audience that never reaches the product.

## A weekly rhythm that can survive a release week

Suppose one person has an afternoon for marketing. They can spend the first hour reviewing replies and deciding whether the buyer definition changed. The next block goes to fresh outreach while the reasons are still clear. The remaining time goes into one lasting asset: a search page, a directory profile or a landing-page revision.

The following week starts with what happened, not with the old calendar. If replies were flat, inspect the list and the timing before writing more copy. If people answered but did not try the product, the handoff to the page needs work. If trials started and disappeared, marketing has done enough; the product now has the problem.

That is a strategy a small team can run. It has a named buyer, a way to learn quickly, a body of work that accumulates and a clear point where more spend becomes sensible. Everything else can wait until one of those pieces earns the right to grow.

## Cold email vs cold calling: which fits your sale?

https://kamugrows.com/blog/cold-email-vs-cold-calling
Published 2026-09-10.

Cold email gives a buyer time to inspect your claim. Cold calling gives you an answer now. Choose from deal shape, urgency, access and operator time.

Cold email vs cold calling comes down to how the buyer needs to understand the sale. Email gives them time to inspect a specific claim, check the product and answer when the timing fits. A call gives you an immediate conversation, including the objections no landing page revealed.

Neither channel repairs a weak list. If the person has no reason to care, email gets ignored and the call gets ended. The useful choice starts with the deal: how urgent the problem is, how much explanation it needs, whether you can identify a trigger and how much operator time you can give each prospect.

## Where cold email works better

Email suits a claim that can be checked without you in the room. You can point to the buyer's recent launch, explain why it made you write and give them a link to inspect. The recipient can forward the message to the person who owns the problem, which matters when the first contact is close to the decision but not responsible for it.

It also works around a founder's calendar. Research and writing happen in batches, replies arrive over several days, and a useful thread can move to a call once there is something to discuss. That makes email a reasonable first channel when nobody can spend a fixed block every morning dialling.

The trade is delayed feedback. Silence can mean the person was wrong, the timing was wrong, the message was unclear or the mail never reached the inbox. You have to separate those possibilities by checking addresses, watching bounces and comparing [reply rates](/glossary/reply-rate) across tightly defined groups.

Cold email also carries sender risk. Authentication, domain age, list quality, volume and complaints all affect whether later messages arrive. A careful sending setup is part of the channel, not an administrative job after the copy is done.

## Where cold calling works better

A call works when the problem is live and the value takes a conversation to uncover. The person can correct your assumption in the first minute, ask how the product handles their odd case and tell you who actually owns the decision. That density of feedback is hard to match in a thread.

Calling is also useful when the buyer rarely handles unsolicited email but answers a business line as part of the job. Local services, operations teams and roles with a public number may fit that pattern. Check the rules that apply to the place and the type of number before you call; permission and do-not-call requirements differ by market.

Operator time is the cost. Research helps, though the call itself cannot be prepared once and delivered later. Someone has to dial, wait, listen and recover after a string of people who were busy. A good caller gets better through those conversations. A founder who avoids the block on the calendar gets no result at all.

## Compare the shape of the sale

| Question | Cold email | Cold calling |
| --- | --- | --- |
| Can the value be understood from a short claim and a link? | Usually a good fit | Often unnecessary |
| Does the buyer need several questions answered before the claim makes sense? | Use it to ask for a conversation | Stronger fit |
| Is there a visible [buying signal](/glossary/buying-signal) you can mention? | Gives the message a reason to exist now | Gives the opener a reason, too |
| Can someone give the channel a fixed daily block? | Work can be batched | The block is required |
| Does the account have several possible owners? | Easy to forward | Faster to ask for the right person |
| Would a missed week stop the whole channel? | Scheduled follow-up can keep moving | Yes, unless another caller takes over |

The table does not pick for you. It exposes the operational fit, which matters because a theoretically strong channel that nobody runs produces nothing.

## Use both when the account earns it

The channels can share research. A specific observation about a company can open an email and give a caller a reason to ask for one person. What should not be shared is a generic sequence that gives every account the same effort.

For a short list of accounts with large deals behind them, send a plain email first and call later with a reference to it. The email gives the buyer something they can find after the call, while the call answers whether the problem exists now. Stop when they decline. A sequence that keeps switching channels after a clear no is persistence in the least flattering sense.

For a larger set of accounts with smaller contracts, the call time may never pay back. Use email, narrow the [ideal customer profile](/glossary/ideal-customer-profile), and reserve calls for people who reply or show a stronger signal. The channel mix should follow expected deal value and evidence of timing.

## Decide with a small test

Pick one buyer group and one reason to contact them. Split the accounts into two comparable sets, then run each channel long enough to get more than a handful of outcomes. Count qualified conversations rather than touches.

Record the reason each conversation moved or stopped. If calls reach people but the problem never feels urgent, the segment may be wrong. If emails arrive and nobody answers, inspect the trigger and the claim before raising volume. If both create conversations and only one creates suitable trials, the handoff after contact decides the choice.

Do not compare a skilled caller with a mail merge, or a researched email with somebody reading a script for the first time. The comparison is between the versions your team can keep running.

For a solo founder, cold email is usually the easier place to start because the work fits around product and support. Calling wins when each account matters enough to justify live time and the sale improves once someone can ask the next question. Choose the one whose ordinary week you can repeat, then use the other only where it earns the extra work.

## What an agent should not be allowed to do

https://kamugrows.com/blog/what-an-agent-should-not-be-allowed-to-do
Published 2026-08-22.

Autonomy is bounded by numbers, not by trust. Where the checkpoints belong when software is doing the work unattended.

The interesting engineering in an agent is not the model. It is the boundary.

Once software chooses its own next step, its mistakes stop being sentences and start being actions. A chatbot that gets something wrong has produced a bad paragraph. An agent that can send email has sent the wrong email, to two hundred people, from your domain, and there is no recalling it.

So the useful question about any product that works unattended is not how clever it is. It is: what is it not allowed to do, and what enforces that?

## A prompt is not a control

The most common answer to "what stops it doing something bad" is a sentence in a system prompt asking it to be careful. That is a preference, not a control. The model can misread it, and when it does there is nothing in the way.

Real controls are structural, and there are only a few kinds:

**Not having the tool at all.** The strongest guarantee about an action is that no code path exists for it. An agent that cannot delete a customer record will not delete one, regardless of what it concludes.

**Limits enforced in code.** A daily cap that lives in a prompt is a suggestion. A daily cap checked before every send is a fact. The distinction sounds pedantic right up to the first time a run decides it has a good reason to exceed it.

**Approval on the actions that matter.** Not everything, but the specific set below that spends money, reaches an audience, or cannot be undone.

**A log with reasons.** Not for compliance theatre. Because the only way to trust unattended work is to be able to check it afterwards, and "what did it do" is a much weaker question than "why did it do that".

## Where the checkpoints go

The instinct is to review everything. That does not work, and it fails in a way that is worse than it looks: the bottleneck moves to the review queue, the queue gets long, and by Friday you are approving without reading. Now you have the appearance of oversight and none of the substance, which is more dangerous than having no checkpoint at all, because you have stopped expecting to be surprised.

The opposite instinct, reviewing nothing, means finding out about a mistake after it has been repeated two hundred times.

The rule that actually holds is to place checkpoints by **reversibility**, not by importance. An important decision you can undo tomorrow is safer to automate than a trivial one you cannot.

In practice that means research, drafting and internal work run alone. Publishing, sending at volume, and spending wait for a person. The second category is deliberately small, which is what keeps the review load survivable and therefore keeps the reviews real.

## What this looks like when it is done properly

Some concrete examples, because the principle is easy to agree with and easy to not implement.

A video ad gets written, cast, rendered and built into a campaign, then left **paused**. Everything expensive is done; starting it is a deliberate human act. The agent did the work and did not take the risk.

Outreach sends inside a daily cap the agent cannot exceed, from a domain with warmup, suppression lists and unsubscribe handling enforced in code rather than requested of the model. A bounce and complaint circuit breaker stops sending automatically, because [a sender reputation](/glossary/sender-reputation) falls in days and recovers over weeks, and by the time a person notices the problem in a dashboard the damage is already done.

Anything that changes the plan or spends money arrives as a decision with the reasoning attached, so the answer is a judgement rather than a guess.

And every action is logged with why, so a wrong one can be found and understood instead of just noticed.

## Be sceptical of "fully autonomous"

Treat any claim of full autonomy as marketing. Reviews of AI SDRs across the category, including the [ones sold at enterprise prices](/compare/artisan), consistently report that teams still spend hours weekly keeping messaging on brand. That is not a failure of those products. It is what the work is.

The honest version of the pitch is not that nobody is needed. It is that the everyday production work happens without you, the small set of consequential calls comes to you, and you can check the rest afterwards because it wrote down what it did and why.

Anyone telling you the number of decisions you have to make is zero is describing a product that has not met a real customer yet.

## The cheapest tool is the one nobody opens

https://kamugrows.com/blog/the-cheapest-tool-is-the-one-nobody-opens
Published 2026-08-18, updated 2026-08-30.

A founder's growth stack usually fails on attention rather than on features. What that means when you are choosing what to buy.

There is a specific way a founder's growth stack dies, and it is not the way the reviews prepare you for.

You buy a good tool. You spend a weekend setting up its fields, sequences and enrichment logic. It works. Then you ship a release, a customer has a problem, an investor wants a call, and three weeks later there is a campaign paused in a tool you are still paying for.

Nothing about that was a product failure. Every one of those tools does what it says. The thing that ran out was not budget or features. It was the one input none of them can supply, which is somebody's attention.

## What the sticker price leaves out

Most comparisons between growth tools are about capability: whose database is bigger, whose personalisation goes deeper, whose deliverability is better. Those are real differences and worth reading about.

They are also all downstream of a question nobody puts in a comparison table: **who is going to operate this?**

A workbench like [Clay](/compare/clay) has a genuinely high ceiling. You can express targeting logic that no automated system would infer, choose which provider answers which field, and audit exactly why a row scored the way it did. That ceiling is only reachable by someone doing the expressing, and it stays reachable only while someone keeps doing it.

A per-seat tool like [Apollo](/compare/apollo) is priced as a workstation, because that is what it is. It is a reasonable price for a person whose day is spent in it. For one founder it is a reasonable price for a room you walk past.

A sequencing tool like [lemlist](/compare/lemlist) can produce outreach that beats anything generated: a real piece of research per account, a custom image that could only have been made for that company. Done in the ten minutes you have on a Tuesday, it produces a template with a first name in it, which is the thing every recipient has learned to ignore.

None of that is a criticism of those products. It is a description of what happens when the scarcest input is not the one being priced.

## Two honest options

Given that, there are really only two things that work.

**Hire someone.** A growth person's whole job is being the attention those tools need. This is the correct answer more often than founders like to admit, and if you can justify the hire, most of the tooling question resolves itself.

**Buy something that does not need you.** The reason [a growth agent](/glossary/growth-agent) is a different category from a marketing tool is not that it has AI in it. Nearly everything does now. It is that the work happens on a schedule of its own, whether or not anyone opens it.

What you give up is control. You do not compose the workflow, which means you cannot tune it the way an operator can. That is a real loss and worth being clear-eyed about: if you have someone who will tune it, the workbench wins and it is not close.

## What to actually ask when you are choosing

Not "which of these is best". Try these instead.

**If I do not open this for three weeks, what happens?** If the answer is "nothing", you are buying a workbench, and you need to know who is standing at it.

**What is the realistic version of the output, not the demo version?** The demo is what the tool does with unlimited attention. The realistic version is what it does with yours.

**What does it cost me when it goes wrong while I am not watching?** This is the question that gets skipped, and it is the one that matters most for anything autonomous. An email to two hundred people cannot be recalled. A draft nobody sent costs nothing. The difference between those two is the whole design problem, and any product doing work unattended should be able to tell you exactly which of its actions are reversible.

## The part that is not a tooling decision

One caveat, because it would be dishonest to leave it out. None of this fixes positioning. If you do not know who buys your product and why, no amount of software will find them for you. It will just find the wrong people faster and more consistently.

That is worth an afternoon with a notebook before it is worth a subscription to anything, including ours.

## Why your cold email lands in spam

https://kamugrows.com/blog/why-your-cold-email-lands-in-spam
Published 2026-08-12, updated 2026-08-30.

It is almost never the words. Six things decide whether your email arrives, and five of them are about you rather than the message.

People who ask why their cold email went to spam usually start by rewriting the subject line. The receiving provider had often made its decision before anyone read a word of it.

A receiving provider decides where to put your message using evidence about **you**: whether the mail is authenticated, what your domain has done before, and how the people who got your last emails behaved. The content is a small input, and the folklore about trigger words is left over from filters that stopped mattering fifteen years ago. Removing the word "free" from an email sent by a domain with no history changes nothing.

Here is what actually decides it, roughly in order of how much damage each one does.

## Your domain is new and you started at volume

A domain that has never sent anything and then sends five hundred emails on a Tuesday looks exactly like a domain bought that morning for spam, because that is very often what it is. Providers have no history to weigh, so they treat the absence of history as risk.

The fix is slow and cannot be rushed: start at a handful a day and roughly double every few days, pausing whenever bounces or complaints move. Three to six weeks to reach a normal outreach volume is the honest number. Sending more does not shorten it, and sending too much lengthens it.

## You are not authenticated properly

SPF, DKIM and DMARC are three DNS records that let a receiver check the mail really came from your domain. Most people set up one or two and stop.

That does not work. SPF alone breaks on any forwarded message and says nothing about the From address a person actually reads, which is the field that gets spoofed. DKIM proves the message was authorised but tells receivers nothing about what to do with failures. DMARC is what makes the other two enforceable, and since 2024 Google and Yahoo require it on any domain sending meaningful volume.

If you set DMARC today, start at `p=none` and read the reports for a few weeks before enforcing. Almost every organisation discovers a legitimate sender it had forgotten about, such as a billing system or help desk, that is not covered by its setup. Going straight to `p=reject` means silently destroying your own mail.

## You are mailing addresses that do not exist

Bounces are the fastest way to damage a domain, and every one is avoidable. Whether an address can receive mail is knowable in advance, and mailing a list you have not checked is precisely what a bought list looks like from the receiving side.

Keep bounces under 2%, and under 1% if you can. Verification costs a fraction of a cent per address. Inbox placement costs weeks to get back.

## People are marking you as spam

Complaints are weighted heavily and the threshold is lower than most people expect: past roughly 0.3%, Gmail starts filtering most of what you send. One campaign to a stale or poorly targeted list can do months of damage.

The counterintuitive part is that making it *easier* to unsubscribe protects you. Someone who cannot find the exit uses the spam button instead, and those two actions do very different amounts of harm.

## Nobody is replying

Engagement is a positive signal, and replies count for more than opens because they are harder to fake. This is why a low reply rate and a deliverability problem are usually the same underlying fact seen from two sides: mail nobody answers looks like mail nobody wanted.

It is also why sending more of a message that is not working makes things worse rather than better. The number to manage is replies per thousand sent, and it moves on relevance and timing, not on volume.

## You are sending from the domain your business runs on

Outreach carries a risk of complaints that ordinary business mail does not, and a damaged domain affects everything sent from it: invoices, password resets, replies to customers. A separate subdomain contains that risk. The cost is that it starts with no reputation and has to be warmed, which is the trade worth making.

## The uncomfortable summary

Five of the six are about your sending setup and your targeting. One is about the message. If you have a deliverability problem, the hour you were going to spend rewriting copy is better spent on your DMARC record and your list.

And if your reply rate is under about 2%, the problem is not the wording either. It is that you are writing to people who have no reason to care this week.

# Comparisons

## A Clay alternative that does the work

https://kamugrows.com/compare/clay

Clay is a go-to-market workbench for building enrichment and outbound workflows. Claims checked 2026-08-27.

Verdict: Clay is the better product if you have someone whose job is to build and tune outbound workflows. Kamu is the better product if you do not, because the work happens without anyone opening it. They are not really the same category: one is a workbench, the other is a worker.

Clay is very good at what it is. You compose a table, pick enrichment providers per column, write the conditional logic, and get precise control over how a prospect list is built and scored. Teams that know exactly what they want get further with Clay than with almost anything else.

That control is also the cost. A Clay workflow produces nothing until someone designs it, and keeps producing only while someone maintains it. The 2026 pricing model makes the same point in a different way: it charges Data Credits for the data and Actions for each step of each workflow, so the shape of what you built shows up on the bill. That is a reasonable trade when a person owns the outbound function. It is the wrong trade when the person who would own it is the one writing the product.

Kamu starts from the other end. You give it a URL, it decides who is worth writing to and what to say, and it does that daily inside hard plan limits. You do not compose the workflow, which means you also cannot tune it the way a Clay operator can. What you get instead is that nothing waits on your attention.

- Who does the work. Kamu: Kamu decides what to do next and does it on its own schedule, unattended. Clay: You design the workflow; Clay runs what you designed, when you run it.
- What you configure. Kamu: Objectives and pace. You choose which jobs are on and how quickly Kamu moves. Clay: Tables, columns, providers, conditional logic, and the sequence that follows.
- Scope. Kamu: Outbound plus SEO articles, social posts, video ads, a landing page and directory listings. Clay: Prospecting, enrichment and outbound. Content and ads are elsewhere.
- What it costs. Kamu: €25 Seed, €49 Starter, €149 Growth, €349 Scale, monthly. No free tier, cancel any month. Clay: Free tier at 500 actions. Launch from $167/mo, Growth from $446/mo, or $54 and $185 billed annually.
- What the bill tracks. Kamu: Credits per action, with a plan ceiling Kamu stops at rather than exceeds. Clay: Data Credits for enrichment plus Actions per workflow step: 15,000 actions on Launch, 40,000 on Growth.
- Trying it first. Kamu: No trial. You pick a plan before it starts. Clay: A free tier that never expires: 500 actions and 100 data credits a month, unlimited seats.
- Who it is built for. Kamu: Solo founders and small teams with nobody doing growth full time. Clay: Go-to-market and sales-ops people, increasingly mid-market and up.

When Clay is the better choice: Pick Clay if someone will own it. If you have a growth or sales-ops person, Clay's ceiling is higher than Kamu's and it is not close: you can express targeting logic Kamu will not infer, plug in your own data providers, and audit exactly why each row scored the way it did. Pick Clay too if enrichment quality is the whole problem you are solving, or if your outbound depends on data sources Kamu does not read. And if you already have a working Clay build, keep it. Kamu is not a migration target for a system that someone is already tending well.

Q: Is Kamu a Clay alternative?
A: Only if what you wanted from Clay was the outcome rather than the control. Clay is a workbench: its value is that you can express precisely how a list is built and scored, and that value is only realised by someone doing the expressing. Kamu is an agent that decides those things itself within hard plan limits. If your Clay account is unused because nobody had time to build in it, Kamu is a real alternative. If it is used well, Kamu is a downgrade in control.

Q: Can Kamu do the enrichment Clay does?
A: Less of it, and less precisely. Kamu finds people who match the buyers in its plan, verifies that the address works before writing, and reads public signals about what someone recently shipped or posted. It does not offer a marketplace of enrichment providers per field, and it will not let you choose which vendor answers which column. If per-field provider control is what you need, that is Clay's job, not Kamu's.

Q: Which is cheaper?
A: It depends on volume and on whose time you are counting. Both publish self-serve prices, so the sticker comparison is easy to make on their pricing pages. The comparison that actually matters is that Clay's output scales with the hours someone puts into it, and those hours are usually the scarcest thing a small team has. A cheap tool nobody operates costs more than an expensive one that runs.

## An Apollo alternative for founder-led sales

https://kamugrows.com/compare/apollo

Apollo is a contact database with sequencing and dialling, priced per seat. Claims checked 2026-08-27.

Verdict: Apollo sells you access to contacts and the tools to work them. Kamu sells you the work. If you have people who will sit in a sequencing tool every day, Apollo is built for them; if you do not, most of what you would pay Apollo for goes unused.

Apollo's core asset is a very large contact database with filters over it, plus the sequencing, dialling and reporting a sales team needs around it. Its pricing is per user per month, with a free tier and published paid tiers, and credits attached to each seat. That model tells you exactly who it is for: an organisation with seats to fill.

The per-seat shape matters more than the headline price. A team of ten pays ten times the sticker, credits do not roll over, and a fully enriched contact with phone data consumes several credits at once. None of that is unreasonable for a sales org, where the seat is a person whose day is spent in the tool. It is a poor fit for one founder, because the unit you are buying is a workstation and you only need the output.

Kamu has no seats in that sense. It finds people who match the buyers in its plan, verifies the addresses, writes each message against something the recipient actually did, sends inside a daily cap, and answers the replies. The comparison Kamu invites is with hiring someone to do that, not with buying them a better place to do it.

- What you are buying. Kamu: The work: prospects found, messages written and sent, replies answered. Apollo: Access to a contact database, plus the tools and seats to work it.
- Pricing shape. Kamu: One plan per account with credits. Cost does not rise with headcount. Apollo: Per user per month. Ten people cost ten times one, and credits sit with the seat.
- Who writes the emails. Kamu: Kamu writes each one, referencing something that person posted or shipped. Apollo: You do, in a sequence. AI assists with drafting inside it.
- Who answers replies. Kamu: Kamu reads the thread and answers on its next run. Apollo: A person, in the inbox or the CRM.
- Scope. Kamu: Outbound plus content, social, ads, landing page, directories. Apollo: Outbound sales: data, sequences, dialler, reporting.
- Who it is built for. Kamu: Founders with no sales team and no time to become one. Apollo: Sales teams, from one rep to an enterprise floor.

When Apollo is the better choice: Pick Apollo if you have people selling, or will soon. A rep who works a list all day wants a database they can filter, a dialler, call recording and a pipeline view, and Apollo gives them all of it in one place at a price that is genuinely reasonable per seat. Its database is also far larger than anything Kamu assembles on its own. If you need coverage of obscure industries, specific geographies or phone numbers, Apollo will find people Kamu will not. And if you already run a CRM-centred sales process, Apollo fits into it; Kamu is not a CRM and does not try to be one.

Q: Is Kamu cheaper than Apollo?
A: For one person, usually not on sticker price alone. Apollo has a free tier and a low entry seat. The difference appears in what you get for it. Apollo's price buys a workstation that produces nothing unless someone sits at it; Kamu's buys work that happens whether or not anyone logs in. For a solo founder the relevant comparison is not two subscriptions, it is a subscription against the hours you would spend in the cheaper one.

Q: Does Kamu have a contact database like Apollo's?
A: No, and this is a real limitation rather than a positioning difference. Apollo licenses and maintains a database of hundreds of millions of contacts with filters over it. Kamu finds people per run against the profile in its plan and verifies each address before writing. That produces a smaller, fresher, more targeted set, and it means there are searches Apollo can answer that Kamu simply cannot.

Q: Can I use both?
A: Yes, and some people should. Apollo as the database and Kamu as the thing that writes and follows up is a coherent arrangement, particularly if you have already paid for Apollo's data. What you should not do is run two systems sending to the same people from the same domain: overlapping outreach doubles the complaint risk against a sending reputation that takes weeks to rebuild.

## A lemlist alternative that runs the outreach

https://kamugrows.com/compare/lemlist

lemlist is a per-seat multichannel sequencing tool with heavy personalisation features. Claims checked 2026-08-24.

Verdict: lemlist is a good sequencing tool for someone who will write the sequences. Its personalisation runs deep, and you build the variables, images and logic yourself. Kamu writes the message instead of giving you better tools to write it.

lemlist's strength is multichannel sequences with unusually deep personalisation: custom images, video, per-recipient variables, LinkedIn steps alongside email. It is priced per seat, and it raised prices in February 2026, with the entry tier starting around $69 per seat per month at the time of checking.

The thing to understand about that personalisation is that it is a set of tools rather than an output. The images, the variables and the branching are things you construct. Done well the results are excellent and better than anything generated; done in the ten minutes a founder has on a Tuesday, they are a template with a first name in it, which is what every recipient has learned to ignore.

Kamu writes each message itself against something the recipient actually did: a post, a launch or a hire. It answers the reply when it comes. You get less control over the shape of the sequence and no image builder. You also do not have to be the one writing at nine on a Tuesday night.

- Who writes the message. Kamu: Kamu, per recipient, from something that person actually posted or shipped. lemlist: You, once, with variables that fill per recipient.
- Personalisation model. Kamu: Generated per person at send time. No image or video builder. lemlist: Manual and deep: custom images, video, liquid-style variables, branching.
- What it costs. Kamu: €25 to €349 a month for the account. Headcount does not change it. lemlist: Email $55/mo billed annually ($69 monthly). Multichannel $87 per user, per month, annually.
- How the price scales. Kamu: It does not, with people. One price per product, whoever is looking at it. lemlist: Per seat on Multichannel, so five people is five times the sticker. Email is unlimited users.
- Trying it first. Kamu: No trial. You pick a plan before it starts, and cancel any month. lemlist: 14 days free on the full Multichannel plan, no card required.
- Sending infrastructure. Kamu: Sends from a pre-warmed domain Kamu owns, or a subdomain of yours once verified. lemlist: You connect your own mailboxes and manage warmup and rotation.
- Beyond email. Kamu: Social posts, SEO articles, video ads, a landing page, directory listings. lemlist: LinkedIn steps and calls inside the sequence. Not content or ads.
- Who it is built for. Kamu: Founders who want the outreach to happen without them. lemlist: Operators who want fine control over a sequence they author.

When lemlist is the better choice: Pick lemlist if you enjoy this work or employ someone who does, and if your outreach is low-volume and high-craft. A genuinely well-built lemlist sequence, with real research per account, a custom image for that company and a well-timed LinkedIn touch, outperforms anything generated. Kamu does not pretend otherwise. Pick lemlist too if you need control over the exact sequence structure and timing, or if sending from your own specific mailboxes with your own warmup regime is a requirement rather than a preference.

Q: Is Kamu a lemlist alternative?
A: It replaces the outcome, not the tool. lemlist assumes you are the author and gives you a very good place to write; Kamu assumes nobody has time to be the author. If your lemlist account has three campaigns in it that were paused in March, Kamu is a real alternative. If you run it weekly and your reply rates are good, changing would cost you control you are actually using.

Q: Does Kamu personalise as well as a hand-built lemlist campaign?
A: Not as well as a good one. A person who researches an account properly and builds something only that company could receive will beat generated copy, and that is worth being straight about. What Kamu beats is the realistic alternative: the template with a merge field that gets sent because there was no time for the good version, or the campaign that never went out at all.

Q: Do I need my own mailboxes and warmup with Kamu?
A: No. Kamu sends from a domain it has already warmed, and you can move to a subdomain of your own once you verify it. Suppression lists, unsubscribe handling, daily caps, warmup and a bounce and complaint circuit breaker are enforced in code rather than left to configuration. With lemlist you connect and manage your own mailboxes, which is more control and more to get wrong.

## An Artisan alternative for smaller SaaS teams

https://kamugrows.com/compare/artisan

Artisan is an AI SDR, sold through a sales conversation rather than self-serve. Claims checked 2026-08-24.

Verdict: Artisan and Kamu agree on the premise: software should do the outbound rather than help you do it. They disagree on who is buying. Artisan is sold through a sales process to companies with a sales function; Kamu is a self-serve subscription for a founder, and covers more than outbound.

Artisan's Ava is an AI SDR: it researches prospects, writes and sends outbound, and works the replies. That is the same shape as Kamu's outreach objective, and where the two overlap they are genuinely comparable.

The difference is how you buy it and what surrounds it. Artisan's pricing page routes most buyers to a sales conversation, and the figures that circulate publicly are estimates from third parties rather than published rates. Reported numbers range widely, which is what happens when a vendor quotes per deal. Reviews consistently mention onboarding periods measured in months and annual commitments. For a company with a sales team and a budget cycle that is normal procurement. For one founder deciding on a Tuesday whether to try something, it is a wall.

The second difference is scope. Artisan is an SDR, so it does outbound. Kamu also writes SEO articles, posts to social, builds video ads and leaves them paused, files directory listings and maintains a landing page. A founder's growth problem is usually not only outbound, and hiring six tools to cover it is its own project.

- How you buy it. Kamu: Self-serve. Published monthly price, start today, cancel whenever. Artisan: Sales-led. Public pricing routes to a request; most tiers are quoted.
- Commitment. Kamu: Monthly, no trial. Yearly is optional and cheaper. Artisan: Reviews report annual contracts as the norm.
- Time to first output. Kamu: Add a URL and it starts working on its own schedule. Artisan: Reviews report onboarding measured in weeks to months.
- Scope. Kamu: Outbound plus content, social, ads, landing page, directories. Artisan: Outbound sales development.
- Who it is built for. Kamu: Solo founders and small SaaS teams. Artisan: Companies with a sales function and a procurement process.
- What is public. Kamu: Prices, plan limits and what each objective does, on the site. Artisan: One self-serve tier is published; the rest is quoted per buyer.

When Artisan is the better choice: Pick Artisan if you have a sales team for it to work alongside. An AI SDR is most useful where there is already a pipeline process, a CRM someone maintains, and people to take the meetings it books. Artisan is built for that context, with the CRM integration and the onboarding support to match. It is also the better choice if you want a named human accountable for making it work, which a sales-led relationship gives you and a self-serve subscription does not. And if outbound is genuinely your only gap, a specialist beats something broader.

Q: Is Kamu an AI SDR?
A: Outreach is one of the six jobs it does, so the SDR comparison is fair for that part and misleading for the rest. An AI SDR is scoped to sales development: prospect, write, follow up, book. Kamu also writes articles aimed at search, posts to social, builds video ads and leaves them paused, and files directory listings. If you want only the SDR part, a dedicated SDR product will go deeper on it.

Q: Why does Artisan not publish its prices?
A: Sales-led pricing is normal for products aimed at companies rather than individuals: the quote depends on volume and use case, and the vendor would rather have the conversation. It is a defensible model. It also means the figures you find online come from third parties, mostly competitors, and vary by an order of magnitude. The only number worth acting on is one Artisan gives you directly.

Q: Which one is more autonomous?
A: Both run without being driven, and both need a person at the same points: anything that spends money, publishes publicly or cannot be recalled. Kamu makes that explicit. Video ad campaigns are built and left paused, and decisions worth judgement go to an inbox with the reasoning attached. Reviews of AI SDRs in general, Artisan included, consistently note that someone still has to watch the messaging weekly to keep it on brand. Treat any claim of full autonomy from either as marketing.

## An 11x alternative without an annual contract

https://kamugrows.com/compare/11x

11x is an enterprise AI SDR sold on annual contracts, with no published pricing. Claims checked 2026-08-24.

Verdict: 11x and Kamu both believe software should do the outbound. They are aimed at opposite ends of the market: 11x is an enterprise purchase with no public price and annual contracts, Kamu is a monthly subscription a founder can start today. If you are one person, the difference in how you buy matters more than any feature.

11x's Alice is an AI SDR that researches prospects and runs outbound across email and LinkedIn. The product has a real following among larger sales organisations, and reviewers credit its onboarding and its targeting.

What is unambiguous is the commercial shape. 11x publishes no pricing and has no self-serve signup: you book a demo. Reported contract values from users and third parties cluster in the thousands per month with annual minimums, and several reviews mention difficulty exiting mid-term. Those figures are estimates rather than list prices. A product you cannot price without a sales call is not a product a founder evaluates on a Tuesday.

Critiques worth taking seriously if you are comparing on capability rather than price: reviewers report that Alice's personalisation leans on static profile data rather than live buying signals, and that teams still spend hours weekly keeping messaging on brand. Kamu writes against recent public activity specifically because timing is most of why outreach works, and it puts a person in the loop at the points that spend money or cannot be undone. Both products need supervision; the honest difference is where.

- How you buy it. Kamu: Self-serve, published monthly price, start immediately. 11x: Demo only. No published price, no self-serve signup.
- Commitment. Kamu: Monthly. Cancel whenever. 11x: Reported as annual, with users describing mid-term exit as difficult.
- Reported cost. Kamu: On the pricing page, in euros, per month. 11x: Not published. Third-party reports vary widely; treat all of them as estimates.
- What drives the message. Kamu: Recent public activity: what someone shipped, posted or hired for. 11x: Reviewers report reliance on static profile data over live signals.
- Scope. Kamu: Outbound plus content, social, ads, landing page, directories. 11x: Outbound sales development, plus a separate voice agent.
- Who it is built for. Kamu: Solo founders and small SaaS teams. 11x: Larger sales organisations with ops support and budget.

When 11x is the better choice: Pick 11x if you are the organisation it is built for. With high-value deals, a sales team to take the meetings, and someone in ops who can own the tool, the economics work in a way they do not for a small team. A sales-led relationship also buys you onboarding help and a person accountable for the outcome. Multichannel coverage across email and LinkedIn under one system, with a voice agent alongside it, is also a real advantage if that is your motion. If your average contract value is large enough that a handful of extra meetings pays for the year, the price stops being the deciding factor.

Q: How much does 11x cost?
A: 11x does not publish pricing and has no self-serve plan, so there is no honest answer that does not come from them. Third-party write-ups and user reports put it in the thousands of dollars a month with an annual minimum, but those are estimates, they vary by a wide margin, and most of the sites publishing them sell competing products. If the number matters to your decision, the only source worth using is a quote.

Q: Is Kamu enterprise-ready?
A: No, and it is not trying to be. There is no SSO, no procurement process, no dedicated onboarding and no account team. It is built for one founder or a small team, priced accordingly, and the things an enterprise buyer needs are genuinely absent. If you are evaluating against 11x because you are an enterprise, that comparison should probably not end with Kamu.

Q: Do either of them work without supervision?
A: Neither works with none. Reviews of 11x report that teams spend hours weekly keeping messaging on brand, and the same is true of Kamu: it writes a standup each morning and puts anything that spends money or cannot be undone into a decision inbox. The difference worth asking about is not whether a person is needed but where they are needed, and how much of the work is reversible if you are not watching.

## An AI growth agent or a growth hire?

https://kamugrows.com/compare/hiring-a-growth-person

A first growth hire is a full-time person who owns marketing and sales for your product. Claims checked 2026-08-27.

Verdict: A growth hire is the better answer if you can afford one and have work that needs judgment rather than volume. Kamu is the better answer if the honest alternative is nobody, which for most solo founders it is. The comparison is not quality against quality; it is a person against an empty chair.

Almost every founder weighing this has already decided marketing needs to happen and is stuck on who does it. The tool comparisons elsewhere on this site are the wrong frame for that question, because the thing being replaced is not software.

A good growth person is better than Kamu at nearly everything that matters. They can read a room, change the strategy because of something a customer said in passing, argue with you, and notice that the whole positioning is wrong. Kamu cannot do any of that. It works a plan it wrote from your site and adjusts it on evidence, which is a much narrower kind of thinking.

What Kamu has is availability and price. It costs from twenty-five euros a month, it starts the day it is switched on, and it does not have a ramp, a notice period or a bad week. A first hire is a real salary, three to six months before they are effective, and a decision that is expensive to reverse. For a product with no revenue yet, those are not small differences.

The honest framing is that these are not competing for the same slot. If you can hire, hire. Kamu is what happens in the eighteen months before you can.

- What it costs. Kamu: From €25 a month on Seed, €49 on Starter. Cancel any month; nothing to unwind. A first growth hire: A salary, employer costs and equity. Varies too much by market and seniority to quote honestly here.
- When it starts working. Kamu: The day you paste your URL. It reads your site and writes a plan on the first run. A first growth hire: Hiring takes weeks, and a good person needs a quarter to understand the product and the market.
- Judgment. Kamu: Narrow. It works its written plan and shifts weight toward what replies. It brings the big calls to you. A first growth hire: The whole reason to hire one. They can tell you the positioning is wrong, which Kamu cannot.
- Volume. Kamu: Up to 200 emails a day on Scale, plus articles, posts and listings, every day, without getting bored. A first growth hire: Whatever one person can do, and the dull half of it is the first thing to slip.
- When it is not working. Kamu: Switch it off. The credits stop and you keep everything it wrote and everyone who replied. A first growth hire: A conversation, a notice period, and a hole where your marketing was.
- Who it answers to. Kamu: You, through a decision inbox for anything that spends money or changes the plan. A first growth hire: You, in a weekly one-to-one, with far more latitude in between.

When A first growth hire is the better choice: Hire the person if you can afford one and the work in front of you needs judgment more than it needs volume. Pricing, positioning, partnerships and a launch that has to land with a specific community all depend on reading a room. Kamu is genuinely bad at that, and a good growth hire will beat it so completely that the comparison is embarrassing. They will also do the thing no software does: tell you that the plan you asked for is the wrong plan. If your product already has revenue and the constraint is scale rather than money, stop reading comparison pages and go hiring.

Q: Is Kamu meant to replace a growth hire?
A: No, and a page claiming otherwise would be lying to you. It is meant to be what happens when the alternative is nobody, which for a solo founder before revenue it usually is. Plenty of customers will hire somebody eventually, and Kamu is a reasonable thing to keep running underneath them: the daily volume work is the half a person is most likely to let slip.

Q: What would a growth hire do that Kamu will not?
A: Change your mind. Kamu reads your site, writes a plan and works it, adjusting toward whatever gets replies. It will not tell you that you are selling to the wrong people, that the pricing is the problem, or that the product needs to change before any of this works. A good hire says all three in their first month, and that is frequently the most useful thing they ever do.

Q: Can I run both?
A: Yes, and it is a sensible arrangement. The person takes the judgment work and the relationships; Kamu takes the daily grind of finding people, writing to them, and keeping the content and listings moving. Every action is logged with the reason behind it, so the person can read what happened rather than being asked to trust it.

Q: What happens to the work if I stop paying?
A: You keep it. The prospects, the replies, the drafts and the published articles are yours and stay in your account. Kamu stops working; nothing is withdrawn.

## An AI growth agent or a fractional CMO?

https://kamugrows.com/compare/fractional-cmo

A fractional CMO is a senior marketer who works with you a few days a month. Claims checked 2026-08-27.

Verdict: A fractional CMO decides what should happen; Kamu does things. That sounds like a hierarchy and is really a division: the common failure of fractional work is a good strategy nobody has time to execute, and the common failure of Kamu is executing steadily in a direction nobody senior has questioned.

Fractional senior marketers exist because the judgment a company needs is not full-time work but the salary is. You buy a few days a month of someone who has done it before, and what you get is a plan, a set of priorities and someone to tell you when you are wrong.

What you do not get is hands. The recurring complaint about fractional arrangements is not the quality of the thinking, it is that the thinking arrives faster than a small team can act on it. A deck lands, the founder agrees with all of it, and three months later two of nine things have happened.

Kamu is the inverse and just as lopsided. It executes daily and indefinitely without needing anybody's attention, and its strategic thinking is genuinely shallow: it reads your site, works out who plausibly buys this, writes that down, and adjusts toward whatever gets replies. It will never tell you the market is wrong.

So the comparison is not really which is better. It is which half you are missing, and a fair number of founders are missing both.

- What you are buying. Kamu: Execution. Prospects found and written to, articles drafted, posts published, listings filed, daily. A fractional CMO: Judgment. Positioning, priorities, channel choice, and someone senior to argue with.
- How it is priced. Kamu: From €25 a month. Each plan has its own hard caps and included credits. A fractional CMO: A monthly retainer for a set number of days. Most do not publish a rate.
- What happens between sessions. Kamu: It keeps working. There are no sessions. A fractional CMO: Nothing, unless somebody on your side is executing the plan.
- Depth of thinking. Kamu: Shallow and written down. The plan is a document you can read and correct. A fractional CMO: The reason to hire one. Years of pattern-matching from companies you have not run.
- Ramp. Kamu: First plan on the first run, same day. A fractional CMO: A discovery period, then a strategy, then execution by somebody.

When A fractional CMO is the better choice: Take the fractional CMO when the question in front of you is what to do rather than whether it is getting done. If you cannot say who your buyer is, why they choose you over the obvious alternative, or which channel should work, no amount of daily execution fixes that. It just gets you to the wrong place faster, which is the specific failure Kamu is capable of. Senior judgment is also the thing that stops you spending a year on a channel that was never going to work for your price point, and a good fractional will tell you that in week two.

Q: Could a fractional CMO direct Kamu?
A: That is arguably the best arrangement available at this size. Kamu keeps a written strategy you can edit, takes instructions in chat or Slack, and logs every action with its reasoning, so a senior marketer can set the direction and then read what actually happened rather than chasing a founder for updates.

Q: Is Kamu strategic at all?
A: A little, and it is honest about the limit. It reads your product and market, writes a plan in plain words, keeps every version, and shifts weight toward whatever is getting replies. What it does not do is question the premise. If the positioning is wrong, Kamu will pursue it diligently.

Q: Which should a pre-revenue founder buy first?
A: Usually the execution, because at that stage the strategy question is often answered by trying things and the constraint is that nobody has time to try them. That is a rule of thumb rather than advice: if you genuinely do not know who buys this, the deck is worth more than the daily sending.

## An AI growth agent or a growth agency?

https://kamugrows.com/compare/growth-agency

A growth agency is an outside team you retain to run marketing for you. Claims checked 2026-08-27.

Verdict: An agency is a team you can hold to account, which is worth real money when there is real money at stake. Kamu is a fraction of the price and shows its work in a way an agency rarely does. The deciding question is usually whether your budget makes you a client anybody senior will actually work on.

Agencies solve a genuine problem: they bring several skills at once, they have run this before, and there is somebody to ring. For a company with budget and a product that already sells, that is often the right purchase.

The trouble at the small end is well known and rarely written down by agencies. A small retainer buys junior time, because the senior people who won the account are working on the accounts that pay. The work gets done to a process rather than to your situation, reporting is monthly and framed favourably, and the thing you most want to know is the thing hardest to get: what was actually tried and why.

Kamu is the opposite trade in every one of those. It is cheap, it is not a team, it has never done this before in any sense that transfers, and it has no relationships to bring. What it does have is a complete log: every email, post, article and listing, each with the reasoning that led to it, visible the moment it happens rather than in a monthly deck.

If your budget puts you at the bottom of an agency's client list, that trade is better than it sounds.

- What it costs. Kamu: From €25 a month, up to €349 on Scale. Published, and the same for everyone. A growth agency: A monthly retainer, usually with a minimum term. Most do not publish a rate.
- Who does the work. Kamu: The agent, every day, inside the plan's hard caps. A growth agency: A team, at the seniority your retainer buys, on their schedule.
- What you can see. Kamu: Every action, with the reason, as it happens. Nothing occurs off the record. A growth agency: A monthly report, written by the people being assessed in it.
- Range of skills. Kamu: Email, SEO articles, social, a landing page, directories and Meta ads. No design, no PR, no events. A growth agency: Whatever the agency staffs. Usually broader, including work Kamu cannot do at all.
- Getting out. Kamu: Cancel any month. You keep the prospects, the drafts and the published work. A growth agency: A notice period, and sometimes an argument about who owns the accounts and the content.

When A growth agency is the better choice: Retain the agency when the work needs people: brand and design, PR, events, partnerships, anything where relationships are the deliverable. Take one too when the stakes justify accountability, such as a real budget going into paid, a launch that has to land on a date, or a category where getting it wrong is expensive. And take one if you want somebody to own the outcome rather than the activity, because that is the thing Kamu genuinely cannot offer. It does the work and shows you exactly what it did; it cannot be held responsible for whether the strategy was right.

Q: Is Kamu cheaper than an agency?
A: By a large multiple at every plan, which is the least interesting thing about the comparison. The question worth asking is what a small retainer actually buys at a given agency, because at the bottom of the range it is often junior time against a template, and that is the version Kamu compares well to. Against a senior team properly resourced, it does not.

Q: Do I keep the work if I leave?
A: Yes. The prospects, the conversations, the drafts and the published articles are in your account and stay there. Cold email sends from your own domain once you verify one, and the social accounts are yours because you connected them, so there is nothing for anybody to hold on to.

Q: Can Kamu work alongside an agency?
A: Yes, and the split that tends to work is the agency taking the campaign and brand work while Kamu keeps the daily volume moving. Objectives are switched on one at a time, so you can hand it exactly the half nobody else wants and leave the rest off.

Q: What can an agency do that Kamu cannot?
A: Design, PR, events, partnerships, anything needing a relationship, and any channel outside the six objectives. It also cannot be accountable in the way a supplier is: it will tell you precisely what it did and why, and it will not tell you whether that was the right thing to have done.

## An Instantly alternative that writes the email too

https://kamugrows.com/compare/instantly

Instantly is a cold email sending platform with mailbox warmup and deliverability tooling. Claims checked 2026-08-27.

Verdict: Instantly is the better product if you know who you are writing to and need to reach a lot of them reliably. Kamu is the better product if the hard part is deciding who to write to and what to say. On raw sending volume per euro the comparison is not close, and it is not close in Instantly's favour.

Instantly is sending infrastructure, and good sending infrastructure is a real thing to buy. Unlimited mailboxes, warmup running continuously across them, deliverability tooling, and on the upper plans dedicated IP blocks. If your problem is that you have ten thousand addresses and need the mail to actually arrive, this is the shape of product that solves it.

It assumes you bring the campaign. Someone has to decide the segment, write the sequence, and judge whether the reply is worth following. Instantly has added AI features on top of that, but the centre of the product is still a platform an operator runs.

Kamu is not sending infrastructure and does badly if judged as though it were. It sends between ten and two hundred emails a day depending on plan, one at a time, each written against something that specific person actually posted or shipped. The pacing is deliberate rather than a limitation of the plumbing: the whole premise is a small number of messages a person might reasonably answer.

So the two are answering different questions. Instantly asks how to reach many people reliably. Kamu asks who is worth writing to this week and what to say to them, and then does it without anybody opening anything.

- What it costs. Kamu: €25 a month on Seed, €49 Starter, €149 Growth, €349 Scale. Instantly: Growth $47, Hypergrowth $97, Lightspeed $358 for outreach; bundles from $94 to $555.
- Sending volume. Kamu: 10 to 200 emails a day by plan, so roughly 300 to 6,000 a month. Instantly wins this outright. Instantly: 5,000 a month on Growth, 125,000 on Hypergrowth, 500,000 on Lightspeed.
- Who decides the target. Kamu: Kamu does. It reads your product, works out who buys it, and finds those people. Instantly: You do. The platform sends the list you bring or build.
- Who writes the message. Kamu: Kamu, one at a time, opening on something that person actually said or shipped. Instantly: You write the sequence; the platform personalises it from your columns.
- Mailboxes and warmup. Kamu: One sending domain, warmed by Kamu. There is no mailbox fleet to manage. Instantly: Unlimited mailboxes and unlimited warmup on every outreach plan. This is the core of it.
- What else it does. Kamu: SEO articles, social posts, a landing page, directory listings and Meta ads. Instantly: Email, and lately AI agents around it. It is not trying to be the rest.

When Instantly is the better choice: Take Instantly when volume is the point. If you have a defined audience, a list already built or bought, and a need to reach thousands of them a month, Kamu is the wrong tool and the pricing makes that obvious: five thousand sends is a month of Instantly's cheapest plan and more than Kamu does at any tier. Instantly also wins decisively on deliverability engineering. A fleet of warmed mailboxes across many domains is a genuinely different discipline from one warmed sending domain, and at scale it is the difference between arriving and not. If somebody on your side owns outbound and is good at it, giving them infrastructure beats giving them an agent.

Q: Why does Kamu send so few emails?
A: Because it writes each one against something specific about that person, and that is the only reason a stranger replies. The cap is a design decision rather than a plumbing limit: a small number of messages somebody might answer is the premise, and thousands of them is a different business with a different deliverability problem.

Q: Could I use both?
A: Yes, and it is a coherent split if you have the volume to justify it. Instantly runs the campaigns somebody on your side designs; Kamu handles the long tail, the content and the channels nobody has time for. They do not conflict, because they are not sending to the same list for the same reason.

Q: Does Kamu do mailbox warmup?
A: It warms its own sending domain and enforces the caps, suppression and bounce handling in code rather than asking a model to remember them. What it does not do is run a fleet of mailboxes across many domains, which is the thing Instantly is built for and the reason it can send at the volumes it does.

Q: Which is cheaper?
A: Kamu, per month, at every tier. Instantly, per email sent, by a very large margin. Which of those matters depends entirely on whether your problem is volume or deciding what to send.

## A Smartlead alternative for lower-volume outreach

https://kamugrows.com/compare/smartlead

Smartlead is a cold email platform with unlimited mailboxes, warmup and a built-in CRM. Claims checked 2026-08-27.

Verdict: Smartlead is cheap, high-volume sending with warmup and a CRM attached, and at $39 a month it is cheaper than Kamu's entry plan while sending twenty times more. What it does not do is decide who to write to or what to say, which is the part Kamu exists for.

Smartlead's entry plan is $39 a month for six thousand sends, two thousand contacts and a CRM, with unlimited mailboxes and warmup from the Pro tier. As sending infrastructure at the small end that is a strong offer, and it is why the product turns up in nearly every list of cold email tools.

The assumption underneath is the same one every platform in this category makes: you have a list, or you know how to get one, and you know what to say to it. The software's job is to send reliably, rotate mailboxes, and keep the replies organised. Judged on that, it does well.

Kamu is judged on something else and would look poor on those terms. It sends ten emails a day on its entry plan against Smartlead's two hundred, and there is no mailbox fleet at all. What it does instead is the part before sending: it reads your product, decides who plausibly buys it, finds those people, checks the addresses work, and writes each message against something that person actually did.

The comparison, then, is not really about email. It is about whether the bottleneck is throughput or judgment, and a founder with no list and no time has the second problem.

- What it costs. Kamu: €25 Seed, €49 Starter, €149 Growth, €349 Scale, billed monthly. Smartlead: Base $39, Pro $94, Unlimited Smart $174, Unlimited Prime $379. About 17% off annually.
- Sending volume. Kamu: 300 to 6,000 a month by plan. Smartlead beats this at every tier including its cheapest. Smartlead: 6,000 a month on Base, 90,000 on Pro, up to 500,000 on Prime.
- Where the list comes from. Kamu: Kamu builds it. It works out the buyer from your site and goes looking for matches. Smartlead: You. Base stores 2,000 contacts and includes 2,000 verified prospect emails.
- Who writes. Kamu: Kamu writes each email, then checks it against a list of AI tells before sending. Smartlead: You write the sequence. The platform handles spintax, personalisation and rotation.
- Trial. Kamu: No free trial. You pick a plan before it starts, and can cancel any month. Smartlead: Free trial, no card required.
- Beyond email. Kamu: Articles, social posts, a landing page, listings and Meta ads, from the same plan. Smartlead: Email and the CRM around it.

When Smartlead is the better choice: Take Smartlead when you already know who you are writing to and want the sending to be somebody else's problem. At $39 for six thousand sends it is cheaper per email than Kamu by a factor nobody should try to argue with, and the free trial means you can find out whether it fits before paying anything, which Kamu does not offer. It is also the better answer for an agency running campaigns for several clients: unlimited mailboxes, rotation and a CRM are exactly that shape of work, and Kamu is built for one product at a time. If your constraint is throughput rather than deciding what to send, this comparison is not close.

Q: Smartlead is cheaper. Why would I pay more for Kamu?
A: For sending, you would not. Smartlead's entry plan is cheaper than Kamu's and sends twenty times more. You would pay more only if you need the work before the send: deciding who is worth writing to, finding them, checking the addresses and writing something specific to each. Kamu also handles articles, posts and listings. If you have that covered, buy the cheaper sender.

Q: Does Kamu have a free trial?
A: No. You choose a plan before it starts working, and can cancel whenever you like. Smartlead's free trial with no card is genuinely the friendlier way in, and it is worth saying so on a page like this.

Q: Can Kamu send from many mailboxes?
A: No. It sends from one domain, either Kamu's pre-warmed one or a subdomain of yours once verified. Mailbox rotation across a fleet is what makes very high volume possible and is the thing Smartlead is built around; it is not something Kamu does or is trying to do.

Q: Which is better for an agency?
A: Smartlead, clearly. Many clients, many mailboxes, rotation and a shared CRM is the exact shape of agency work. Kamu is built around a single product, reads that product's own site, and writes in that founder's voice, which does not divide across a client roster.

## Kamu vs Soleur, two different growth agents

https://kamugrows.com/compare/soleur

Soleur is a company-as-a-service platform of agents across eight business departments. Claims checked 2026-08-27.

Verdict: Soleur is broader than Kamu by a wide margin and free if you self-host, which are two very good reasons to take it. Kamu is narrower and costs money, and the thing it buys is that the work happens without you approving each piece. Whether that is worth paying for depends entirely on how much you want to review.

Soleur describes itself as company-as-a-service: sixty-plus agents across engineering, marketing, legal, finance, operations, product, sales and support, aimed squarely at the solo founder wearing nine hats. The scope is far wider than Kamu's, which does six things and none of them are legal or finance.

It is also free to run yourself. The self-hosted version is source-available under BSL 1.1, and at time of writing the managed cloud is listed as coming soon with no published prices. For a founder with no budget and some tolerance for infrastructure, that is a genuinely strong offer and Kamu has no answer to it.

The difference that actually decides this is not scope or price, it is the approval model. Soleur is explicit that it is human-in-the-loop by design: every agent output is a starting point, and you review, edit and approve before anything ships. Kamu is built on the opposite premise. It sends the cold email, publishes the post and files the listing on its own schedule, inside hard limits, and brings you only the decisions that spend money or change the plan.

That is the whole comparison. One assumes you want many agents working with your judgment on each output; the other assumes the reviewing is the bottleneck you were trying to remove.

- Scope. Kamu: Six jobs: outreach, SEO articles, social, a landing page, directory listings, Meta ads. Soleur: 60+ agents across eight departments, including engineering, legal and finance.
- Who presses send. Kamu: Kamu does, inside hard caps. Decisions that spend money or change the plan come to you. Soleur: You do. Every output is explicitly a starting point for you to review and approve.
- What it costs. Kamu: €25 to €349 a month, published, hosted. No free tier. Soleur: Free to self-host under BSL 1.1. Managed cloud is listed as coming soon, pricing unpublished.
- What you run. Kamu: Nothing. It is hosted, and there is no infrastructure on your side. Soleur: Your own deployment, if you take the free path. That is the trade for the price.
- Sending and publishing. Kamu: Sends cold email from your domain, posts to connected accounts, files listings itself. Soleur: Produces the work; shipping it is on the other side of your approval.
- Where the knowledge lives. Kamu: A written strategy and memory documents you can open, edit and correct. Soleur: A shared knowledge base the agents compound into, per their own description.

When Soleur is the better choice: Take Soleur if you have no budget, or if the breadth is the point. Sixty agents across eight departments covers work Kamu does not touch at all. Nothing here writes your contracts, does your bookkeeping or reviews your code. A founder who wants help across the whole company rather than depth in growth is asking for what Soleur is. Self-hosting for nothing is also simply a better deal than paying, if you are comfortable running it. And take it if you want to approve everything: plenty of founders do, particularly early, and Kamu is genuinely the wrong shape for somebody who wants to read every email before it goes.

Q: Is Soleur free?
A: The self-hosted version is source-available under BSL 1.1 and free to run, which means you provide the infrastructure. Their managed cloud was listed as coming soon with no published pricing when this page was checked, so a like-for-like hosted comparison is not possible yet.

Q: What is the real difference?
A: Approval. Soleur says plainly that every agent output is a starting point and you review, edit and approve before anything ships. Kamu sends and publishes on its own inside hard plan limits, and only brings you decisions that spend money or change the plan. If reviewing each piece is the bottleneck you were trying to remove, that is the whole comparison; if it is the control you wanted, Soleur is right and Kamu is not.

Q: Soleur does far more. Why would I take the narrower one?
A: Because breadth and depth are a trade, and outreach in particular is mostly unglamorous specifics: checking an address resolves before sending, telling a no from an unsubscribe, spacing follow-ups, keeping off a suppression list across every product you run. Kamu enforces those in code rather than asking a model to remember them. Whether that depth is worth giving up seven departments is a real question, and for a founder whose problem is customers rather than contracts the answer is often yes.

Q: Can I run both?
A: Nothing stops you, and they overlap least where each is strongest. The honest caution is cost and attention rather than conflict: two systems producing work is two systems to keep an eye on, and the reason for either was that you had too little attention to begin with.

## A Hootsuite alternative with no calendar to fill

https://kamugrows.com/compare/hootsuite

Hootsuite is a social media management platform with a calendar, an inbox, listening and analytics for teams. Claims checked 2026-09-03.

Verdict: Hootsuite is the better product for a social team running many accounts out of one inbox, with listening, reports and approvals around it. Kamu is the better product for a founder with one product and no time to fill a calendar, because it writes the posts from what shipped and publishes them itself.

Hootsuite's pitch is one connected place for a social team. Every account sits in one calendar, every comment and message lands in one inbox, and the team can watch what competitors post and how last month's numbers came out. On the Advanced plan a post goes through review and approval before it leaves, and messages are assigned and routed to whoever should answer. That is a lot of product, and it is aimed at a person whose job is social media.

A solo founder does not have that job. They have a product to ship and a feed that goes quiet for three weeks at a time. Hootsuite would give them a calendar with empty slots in it and an AI that drafts posts from trends, on a Standard plan at $99 a user a month billed annually as of September 2026 that covers ten social accounts they do not have. The calendar still needs filling. The tool is very good at publishing whatever you put in it and has no opinion about whether anything is worth saying.

Kamu is built around that gap. It connects to GitHub and reads the releases and merged pull requests, then writes a post for X, a longer one for LinkedIn and, with a picture in your colours, one for Instagram. If nothing shipped that week it says so and stays quiet. Connected accounts get the post published; anything unconnected gets a draft to paste. There is no calendar, because there is nothing to drag into it. There is also no inbox and no listening, which is exactly what a team would miss.

The comparison is between a control room for a social team and a worker for one founder. Which one you need depends on which of those you are.

- What it is. Kamu: An agent that decides what to post, writes it and publishes it, as one part of a wider growth plan. Hootsuite: A social media management platform: calendar, inbox, listening, analytics and approvals in one place.
- Where the posts come from. Kamu: Your GitHub releases and merged pull requests. On a week with nothing shipped, Kamu posts nothing. Hootsuite: You or your team write them, or Hootsuite's AI drafts them from trends, and you schedule them into the calendar.
- Who it is built for. Kamu: A founder with one product and nobody doing social full time. Hootsuite: Social teams, agencies and enterprises with many accounts and a person or a team in the seat.
- Channels. Kamu: X, LinkedIn and Instagram published directly. Reddit comments come to you as drafts. Hootsuite: Up to 10 social accounts on Standard, unlimited on Professional and above.
- What it costs. Kamu: €25 Seed, €49 Starter, €149 Growth, €349 Scale, a month. The daily post cap rises with the plan, from 2 X posts a day on Seed to 25 on Scale. Hootsuite: Standard from $99 a user a month billed annually as of September 2026. The page's FAQ says plans range up to $399 for Advanced. Enterprise is contact sales.
- Trying it. Kamu: No trial. You pick a plan and it starts on your site the same day; the first month on Seed comes with double credits. Hootsuite: A 14-day free trial with no credit card. There is no free plan.
- Inbox and listening. Kamu: Kamu answers people who mention you on X and drafts the replies X will not let it send. It has no inbox, and it does not watch competitors or trends. Hootsuite: One inbox for public and private messages, automated replies and routing on higher plans, and monitoring of your brand and competitors.
- Approvals and reporting. Kamu: You approve anything that spends money or changes the plan. Posts inside the caps go out on their own, each logged with its reason. There is no social analytics dashboard. Hootsuite: Review and approve content with your team on Advanced; custom performance reports on Professional and above.

When Hootsuite is the better choice: Pick Hootsuite if social media is somebody's job. A team with a dozen accounts across brands needs one calendar and one inbox, and needs to know who answered which message; Kamu has neither and will not grow them. Listening is another reason. Hootsuite tracks your brand, your competitors and what is trending, and Kamu reads your own repository, your X mentions and the subreddits you approve, with no view of anyone else. If posts must be reviewed before they go out, the Advanced plan has that built in, and Kamu publishes on its own within its caps. Agencies should not look twice at Kamu: it runs one product per plan on Seed and Starter, up to ten on Scale, and it writes what shipped rather than what a client briefed. And if your team already lives in Hootsuite and the calendar is full, you have a working system. Keep it.

Q: Is Kamu a Hootsuite alternative?
A: For a founder, yes; for a social team, no. Hootsuite is a place to run many accounts, with an inbox, listening and reports around the calendar, and Kamu has none of that machinery. What it has is the part Hootsuite leaves to you: deciding what to say and writing it, from what your product actually shipped, then publishing to X, LinkedIn and Instagram on its own.

Q: Can Kamu schedule posts?
A: Not in the calendar sense. Kamu posts when there is something to post, up to a hard daily cap that depends on the plan: 2 X posts a day on Seed, 5 on Starter, 10 on Growth and 25 on Scale. There is no queue to fill and no slot to drag a post into. If you need a post to go out at nine on Tuesday, Hootsuite is the right tool for that.

Q: Does Kamu reply to comments and messages?
A: Only in a narrow way. It answers people who mention you on X, and for threads where X's rules refuse an automated reply it writes the reply and hands it to you to paste. It has no inbox for Instagram or LinkedIn comments, and nothing to route a message to a colleague. A team that answers customers on social needs Hootsuite or something like it.

Q: Can I use Hootsuite and Kamu together?
A: Yes, and some teams will want to. Kamu publishes directly to the accounts you connect, so there is no queue for Hootsuite to hold. If your team keeps Hootsuite for the inbox and the reports, leave X, LinkedIn and Instagram unconnected in Kamu and it hands you finished drafts instead, which you can schedule from Hootsuite's calendar like anything else.

## A Buffer alternative that fills its own queue

https://kamugrows.com/compare/buffer

Buffer is a low-cost social media scheduler with a free plan, a queue and an AI writing assistant. Claims checked 2026-09-03.

Verdict: Buffer is the cheapest sensible way for one person to schedule posts to many networks, and its free plan is genuinely free. Kamu costs more and does the part Buffer leaves to you: it writes the posts from what your product shipped and publishes them. Buffer holds a queue of posts you wrote. Kamu writes them.

Buffer starts at nothing. The free plan connects three channels, holds ten scheduled posts per channel, includes the AI Assistant and never expires. Essentials is $5 a month for one channel as of September 2026, or $60 a year, and Team is $10 a month for one channel with unlimited members and approval workflows. For a solo founder that is about as cheap as software gets, and it is why Buffer is the default answer when someone asks how to post to X, LinkedIn, Instagram, Threads, Bluesky and Mastodon from one place.

The thing it does not do is write. Buffer's AI Assistant will brainstorm and rewrite, and the queue publishes on time. The queue is still empty on Monday morning until you sit down and fill it. That is fine for a creator whose work is the posting. A founder's work is the product, and the queue is what gets skipped in a busy week.

Kamu costs more, from €25 a month on Seed, and it does not ask you to fill anything. It connects to GitHub, reads what shipped from your releases and merged pull requests, and writes a post for each of X, LinkedIn and Instagram, the Instagram one with a picture it makes in your colours. Then it publishes to whichever of those you have connected and hands you a draft for the rest. On a week where nothing shipped it says so and writes nothing.

You lose a great deal in the trade. Kamu publishes to three networks to Buffer's eleven, and drafts Reddit comments for you to post. It has no queue to reorder and no link in bio page, and Buffer's hashtag manager and follower analytics have no equivalent in it. What you gain is that the posting happens whether or not you had time this week, which for one person shipping alone is usually the whole problem.

- What it is. Kamu: A growth agent. Posting is one of its jobs, alongside cold email, articles, a landing page and directory listings. Buffer: A social media scheduler with a queue, an AI writing assistant, a community inbox and analytics.
- Who fills the queue. Kamu: Kamu does, from your releases and merged pull requests. When nothing shipped, it posts nothing. Buffer: You do. The AI Assistant can brainstorm and rewrite, and you pick the time slot.
- Channels. Kamu: X, LinkedIn and Instagram, published directly. Reddit is drafted for you to post yourself. Buffer: Eleven: Facebook, Instagram, TikTok, LinkedIn, Threads, Bluesky, YouTube Shorts, Pinterest, Google Business, Mastodon and X.
- What it costs. Kamu: €25 a month on Seed for one product and 2 X posts a day, then €49 Starter, €149 Growth and €349 Scale. Buffer: Free for 3 channels and 10 queued posts each. Essentials $5 a month per channel and Team $10 a month per channel as of September 2026, with channels above 10 priced lower.
- Free to try. Kamu: No free plan and no trial. Seed's first month grants twice the usual credits. Buffer: The free plan never expires, and paid plans carry a 14-day trial.
- Working with others. Kamu: Built for one person. You approve anything that spends money; posts within the caps go out on their own. Buffer: The Team plan has unlimited members, approval workflows and custom permissions.
- What you see afterwards. Kamu: A morning standup in chat, Slack, Telegram or WhatsApp, and a log of every post with the reason it was written. Buffer: Insights with 30 days of history on Free, advanced analytics on paid plans, and a community inbox for replies.

When Buffer is the better choice: Pick Buffer if you already know what you want to say and only need it to go out on time. The queue, the free plan and the eleven channels are the right shape for a creator, a small shop or anyone whose posting is part of the job rather than a chore squeezed around it. Pick it too if your audience lives on TikTok, Threads, Bluesky, Pinterest or YouTube Shorts, none of which Kamu touches, or if you want a link in bio page and a first comment scheduled under each post. And if price is the whole question, Buffer wins outright. Its free plan does real work for nothing, and €25 a month for Kamu is only a good deal if the posts it writes would otherwise not exist. For a founder who was going to write them anyway, Buffer is the cheaper and the better tool.

Q: Is Kamu a Buffer alternative?
A: For a founder who never gets round to filling the queue, yes. Buffer is a scheduler: you write, it publishes on time, and the free plan makes that cost nothing. Kamu writes the posts itself from what your product shipped and publishes them to X, LinkedIn and Instagram, from €25 a month. If you enjoy writing your own posts and only need the timing handled, stay with Buffer.

Q: Can I connect Kamu to Buffer?
A: No, and it would not gain you much. Kamu publishes directly to the accounts you connect. If you would rather keep Buffer as the place everything goes out from, leave those accounts unconnected in Kamu: it then writes each post and hands it to you as a draft, which you can paste into Buffer's queue at whatever time you like.

Q: Does Kamu have a free plan like Buffer?
A: No. The cheapest plan is Seed at €25 a month, which runs one product with a cap of 2 X posts a day and one linked LinkedIn or Instagram account, and grants double credits in its first month. Buffer's free plan is a better place to start if you are unsure whether you want to post at all. Kamu makes sense once you know you want the posts and know you will not write them.

Q: Which is better for Instagram?
A: Buffer, if you make your own pictures and want a queue of them. Kamu posts to Instagram with a picture it generates in your brand colours, which spends credits, and it drives a personal account the way a phone app would, with low daily caps. That suits a founder who would otherwise never post there. It is no replacement for a considered visual feed.

## A Surfer SEO alternative that picks the phrase too

https://kamugrows.com/compare/surfer-seo

Surfer SEO is an SEO content platform that scores and optimises drafts against the pages that rank. Claims checked 2026-09-03.

Verdict: Surfer is the better tool for a content team that already writes and wants each draft scored against what ranks. Kamu is the better tool for a founder with no writer, because it chooses the phrase and writes the whole article, then publishes it to Ghost or WordPress once approved.

Surfer works on a draft. You open the Content Editor with a keyword, it reads the pages ranking for it and turns them into live guidelines: length, headings, the terms to include and how often. As you write, a Content Score climbs. Around that sit keyword research, a SERP analyser, a Content Audit that finds weak pages across a site, and an AI writer that drafts to the same guidelines. It plugs into Google Docs and WordPress, and it now tracks how a brand shows up in Google's AI Overviews, ChatGPT, Gemini, Claude and Perplexity. The pricing page, viewed from Finland in September 2026, lists Discovery at €49 a month billed annually for 120 documents and one seat, Standard at €99 for 360 documents and three seats, Pro at €182 for five seats, and Enterprise as a conversation with a Surfer expert.

All of that assumes someone is writing. Surfer's users are marketing managers, SEO specialists, content writers and agencies, and its output is a better version of the article they were going to write anyway. It leaves the choice of what to write to you and makes what you write closer to the top of the page.

Kamu starts before the draft exists. It looks at what your buyers type into Google, weighs search volume against how hard the ranking pages are to beat and whether the person typing would buy, and picks one phrase. It reads what already ranks for it so the draft can be better than that, checks what you shipped recently so the article can mention it, and writes the whole piece with headings, links and a conclusion. Every draft goes past about 120 banned phrases before you see it. You read it and approve; with Ghost or WordPress connected, one button puts it live. The pace is about an article a week, on purpose.

Kamu does not score your own writing, and it does not audit your site. If you have a writer, Surfer is the one to buy.

- What it is. Kamu: An agent that picks the phrase, writes the article and publishes it after approval, as one job among several. Surfer SEO: An SEO content platform: an editor that scores drafts against the SERP, plus research, audit and AI visibility tracking.
- Who writes. Kamu: Kamu writes the whole draft. You read it and press publish. Surfer SEO: Your writer, in the Content Editor or Google Docs, with Surfer's guidelines beside the text; or Surfer's AI writer to the same guidelines.
- Choosing the phrase. Kamu: Kamu shortlists phrases your buyers search, weighs volume against difficulty and buying intent, and picks one per article, with the reasoning kept. Surfer SEO: Keyword research gives you clusters and volumes; you decide which to brief.
- Publishing. Kamu: Connect Ghost or WordPress and an approved draft goes live from one button. Without a CMS you copy the markdown out. Surfer SEO: WordPress and Google Docs integrations on Standard and above; Contentful on Pro.
- What it costs. Kamu: €25 Seed, €49 Starter, €149 Growth, €349 Scale a month. Content drafts are capped per week by plan: 1 on Seed, 5 on Starter, 10 on Growth, 30 on Scale. Surfer SEO: Discovery from €49 a month billed annually as of September 2026, for 120 documents and 1 seat. Standard €99, Pro €182, Peace of Mind €299, Enterprise on request.
- Existing pages. Kamu: Kamu revisits the articles it wrote and refreshes the ones that need it. It does not audit pages it did not write. Surfer SEO: Content Audit finds weak pages across the whole site and tells you what to fix.
- Seats and teams. Kamu: One founder. Kamu is steered in chat, Slack, Telegram or WhatsApp, and every keyword carries the reasoning that chose it. Surfer SEO: 1 seat on Discovery, 3 on Standard, 5 on Pro, 10 on Peace of Mind, with brand spaces for agencies.
- AI search. Kamu: Kamu does not track how you appear in AI answers. Surfer SEO: Tracks visibility across Google AI Overviews, ChatGPT, Gemini, Claude and Perplexity, with prompt tracking on higher plans.

When Surfer SEO is the better choice: Buy Surfer if you have someone who writes. A content team, or a founder who likes writing and has the hours for it, will get more from a scored editor than from an agent that writes for them, because the score makes every piece they were going to write anyway measurably closer to the page that ranks. A large site with old pages is another case for Surfer: the Content Audit finds the weak ones, and Kamu has no view of pages it did not write. So is needing to know how your brand shows up in ChatGPT or AI Overviews, which Surfer tracks and Kamu does not. If your bottleneck is quality rather than output, Surfer is the right side of the trade. Kamu's article a week is a floor for a site with nothing on it, and a ceiling for a team that publishes daily.

Q: Is Kamu a Surfer SEO alternative?
A: Only for a founder who has nobody to write. Surfer improves a draft a person is writing, and it does that well. Kamu removes the person from the drafting: it picks the phrase from search volume and a real read of the ranking pages, writes the article, and publishes it to Ghost or WordPress after you approve. If a writer is already in the seat, Surfer is the tool that helps them, and Kamu is a different purchase.

Q: Does Kamu give a content score?
A: No. There is no number to push up as you type, because Kamu does not expect you to type. It reads the pages that rank for the chosen phrase and writes a draft meant to be better than them, then checks the draft against about 120 banned phrases and a set of shape rules. If you want to grade your own writing against the SERP, that is Surfer's job.

Q: How does Kamu choose what to write about?
A: It looks at the phrases your buyers search around your product, the volume behind each, how hard the ranking pages are to beat, and whether the person typing the phrase would buy. It skips the big vague terms it has no chance at. Each keyword is saved with its intent, its difficulty and the reason it was picked, so you can read the choice rather than take it on trust.

Q: Can Kamu publish to WordPress like Surfer?
A: Yes, and to Ghost. Connect either and an approved draft goes live from one button. Without a CMS connected, the draft stays in Kamu as markdown you copy out. Surfer's WordPress integration moves a draft your writer made into your site; Kamu's publishes an article it wrote itself, after you have read it.

## A Jasper AI alternative that decides what to write

https://kamugrows.com/compare/jasper-ai

Jasper is an AI marketing workspace with brand voice controls, agents and workflows for marketing teams. Claims checked 2026-09-03.

Verdict: Jasper is the better buy for a marketing team that produces a lot of copy and needs every piece to sound like the brand. Kamu is the better buy for a founder with no marketing team, because it decides what to write, writes it and publishes or sends it without anyone sitting at a workspace.

Jasper calls itself the agent workspace built for modern marketing teams, and the phrase is accurate. The centre of it is Jasper IQ, a layer holding brand voices, style guides, audience profiles and product knowledge, so that a hundred pieces of copy from six people all sound like one company. Around it sit more than a hundred specialised agents for jobs like research and translation, a Canvas for working on a piece, a Studio for building custom apps and workflows, Grid for running content automations at volume, image generation, and tracking of how a brand shows up in AI answer engines. Pro is $69 a month for one seat, or $59 billed yearly, as of September 2026, with 2 brand voices, 5 knowledge assets, 3 audiences and a 7-day trial. Business is custom pricing and adds the agent builder, Grid, unlimited voices, API access and SSO.

Every one of those features assumes a marketer at the keyboard, deciding what needs writing and asking Jasper for it. That is the right design for a team with a calendar of campaigns. It is the wrong one for a founder who does not know what should be written this week and has no time to find out.

Kamu makes that decision. It reads your site and your market, writes a plan, and then each day picks work from it: a cold email to a person who fits your buyer, an article for a phrase people search when they want to buy, a post about what shipped this week, a listing in a directory worth being in. It writes the thing and, within its caps and your approvals, sends or publishes it. Every action is logged with the reason.

There is no brand voice library in Kamu; it reads how you already write and works from that. There is no library of agents to choose from and no workflow builder. It has one job, which is to keep your growth moving while you build.

- What it is. Kamu: An autonomous growth agent that chooses the work, does it and reports back. Writing is most of the work, and none of it waits for a prompt. Jasper: An AI marketing workspace: brand voice, agents, a canvas and workflow tools for a team producing copy.
- Who decides what gets written. Kamu: Kamu, from its plan and what changed this week. You steer it in chat and approve anything that spends money or changes the plan. Jasper: The marketer. Jasper produces what it is asked for, in the shape and voice it was given.
- Brand voice. Kamu: Kamu reads your site and how you already write, and writes in that voice. There is no library to maintain. Jasper: Jasper IQ holds brand voices, style guides, audiences and knowledge: 2 voices on Pro, unlimited on Business.
- Where the output goes. Kamu: Out the door: cold email sent from your domain, articles published to Ghost or WordPress after approval, posts published to X, LinkedIn and Instagram. Jasper: Copy, images and campaign assets in the workspace, with a browser extension for writing inside other tools.
- What it costs. Kamu: €25 Seed, €49 Starter, €149 Growth, €349 Scale a month, each with hard caps on emails, posts and articles. Jasper: Pro $69 a month, or $59 billed yearly, per seat, as of September 2026. Business is custom pricing through sales.
- Trying it. Kamu: No trial. The first month on Seed grants double credits, and you can cancel any month. Jasper: A 7-day free trial on Pro.
- Who it is for. Kamu: A founder shipping alone, or a team of two with nobody on marketing. Jasper: Product, content, performance and brand marketers, and enterprises that need admin controls, SSO and SOC 2.
- Scope beyond copy. Kamu: Finding and verifying prospects, a landing page it measures and rewrites, directory listings, and video ads with a generated presenter built paused in Meta. Jasper: Image generation, translation, GEO monitoring, and custom agents and automations at scale on Business.

When Jasper is the better choice: Buy Jasper if you have marketers. A team producing landing pages, ads, emails and posts across several products or markets needs the copy to sound like one company, and Jasper IQ is built for exactly that, with style guides and audiences a whole team draws on. Kamu writes in one founder's voice for one product per plan on Seed and Starter, and has no way to hold a house style for six people. Volume is the second reason: Grid and the agent builder on Business exist to run the same workflow across hundreds of pieces, and Kamu's caps are deliberately low. Enterprise controls are the third, since SSO, SOC 2, admin groups and an account manager are all on Jasper's Business plan and none of them is anywhere in Kamu. And if you want to stay in charge of every word, Jasper never sends anything on its own, and Kamu, within its caps, does.

Q: Is Kamu a Jasper AI alternative?
A: Only if what you want from Jasper is the outcome rather than the workspace. Jasper helps a marketer write faster and more consistently; somebody still opens it, asks for a piece and takes the result somewhere. Kamu decides what to write from a plan it made, writes it, and sends or publishes it inside hard caps. A founder who has no marketer gets the work done. A marketing team gets a downgrade in control.

Q: Does Kamu learn my brand voice?
A: It reads your site and your existing writing and works from that, and it checks every draft against about 120 banned phrases before it goes anywhere. There is no library of voices and style guides to configure, which is a loss for a team and a relief for one founder. If you have a documented brand voice that six people must follow, that is Jasper's job.

Q: Can Kamu write ads and landing pages like Jasper?
A: It writes and publishes one landing page of its own, on its own address, and rewrites it when the visit and click numbers say to. For ads, it writes a short video with a generated presenter and builds the Meta campaign paused in your ad account, or drafts it for you to run by hand. It will not write twenty ad variants against a brief; it makes one thing and measures it.

Q: Which is cheaper?
A: Jasper Pro is $69 a month for one seat, or $59 billed yearly, as of September 2026. Kamu starts at €25 a month on Seed and €49 on Starter. The sticker prices are close enough that the question is whose time you are counting: Jasper saves a marketer hours, and Kamu removes the need for the marketer to be there at all.

## A Copy AI alternative with nothing to build

https://kamugrows.com/compare/copy-ai

Copy.ai is a go-to-market AI platform of workflows, agents and tables that sales and marketing teams configure. Claims checked 2026-09-03.

Verdict: Copy.ai is the better fit for a go-to-market team that wants to build its own workflows on its own data and give each rep and marketer a seat. Kamu is the better fit for one founder who wants none of that built, because it is a single worker that decides what to do and does it.

The name still says copywriting. As of September 2026 the homepage calls Copy.ai the first AI-native GTM platform, and the product is a set of parts a team assembles: Workflows that codify a process, Actions as the building blocks inside them, Agents that run targeted tasks with guardrails, Tables that pull data from several systems into one place, a Brand Voice and an Infobase of company facts. The use cases it lists are prospecting, inbound lead processing, deal coaching, content creation, account-based marketing, translation and CRM enrichment. It runs on OpenAI, Anthropic, Gemini and Perplexity models, and it names Siemens and ServiceNow among its customers.

The pricing follows the shape. Chat is $29 a month, or $24 billed annually, for five seats and unlimited words in chat. Workflow credits begin on Growth at $1,000 a month billed annually, for 75 seats and 20,000 credits, then Expansion and Scale at $2,000 and $3,000, and Enterprise is custom. There is no free plan on the pricing page.

That is a platform for a revenue operations team with data to connect and processes worth codifying. The work is real, and someone has to do it: decide which workflow to build, map the fields, set the guardrails, then keep it running as the CRM changes.

Kamu has no builder. You give it a product URL and it reads the site and the market, writes a plan, and then works the plan every day: finding people who match the buyer, verifying their address, writing to them one at a time, writing an article for a phrase with buyers behind it, posting about what shipped, filing directory listings. You steer it in chat or Slack and approve the calls that spend money. There is nothing to configure beyond which of those jobs are on and how fast it should go.

Copy.ai is a kit for a team to build with. Kamu is one worker for one founder.

- What it is. Kamu: One agent that runs growth for one product: finds buyers, writes and sends, publishes, and reports each morning. Copy.ai: A GTM platform: workflows, actions, agents, tables, brand voice and an infobase a team assembles.
- Who builds it. Kamu: Nobody. Kamu reads your site, writes its own plan and works it. You switch jobs on and off and set the pace. Copy.ai: Your team. Workflows and agents are configured by you, on your data, with your guardrails.
- Who it is for. Kamu: A solo founder or a two-person team with nobody on growth. Copy.ai: Go-to-market teams in sales, marketing and operations; the plans count seats in the dozens and hundreds.
- Outbound. Kamu: Kamu finds people who fit, verifies the address, writes each email itself and drafts the reply when someone answers, inside a daily cap from 10 emails on Seed to 200 on Scale. Copy.ai: Prospecting, inbound lead processing and CRM enrichment as workflows you build and connect to your systems.
- What it costs. Kamu: €25 Seed, €49 Starter, €149 Growth, €349 Scale a month. There are no seats; the plan buys work rather than people. Copy.ai: Chat $29 a month or $24 billed annually as of September 2026. Workflows start on Growth at $1,000 a month billed annually. Enterprise is custom.
- Data and models. Kamu: Kamu brings its own data sources and its own model. You connect email, GitHub, a CMS and social accounts; there is nothing to map. Copy.ai: Tables consolidate your sources; you pick from OpenAI, Anthropic, Gemini and Perplexity models.
- What is refused. Kamu: Unsubscribes are honoured across every product in the account, addresses are verified before sending, campaigns are created paused, and about 120 banned phrases are checked before anything publishes. Copy.ai: The homepage describes guardrails you set on your own agents.
- What you see afterwards. Kamu: A short standup every morning, a decision inbox for anything that spends, and a log of every action with the reason. Copy.ai: Workflow runs and the tables they write to, inside the platform.

When Copy.ai is the better choice: Choose Copy.ai if you have a team and a process worth encoding. A sales team with a CRM full of accounts, or a RevOps person who knows exactly which fields should be enriched from where, has a job for workflows they design, and Copy.ai supplies the parts and the seats to do it. Kamu will not read your CRM or let you design a workflow, and it runs one product per plan on Seed and Starter. Choosing your own model is another reason, since Copy.ai lets you run OpenAI, Anthropic, Gemini or Perplexity and Kamu picks for you. So is procurement: an enterprise plan with a dedicated support team and twenty-plus integrations is something a larger company can buy, and Kamu has no such tier. If your company already has people whose job is go-to-market, a platform that makes them faster is the right purchase. Kamu is for the company that has not got those people yet.

Q: Is Kamu a Copy AI alternative?
A: For a founder, yes; for a GTM team, no. Copy.ai sells the parts of a go-to-market machine, workflows, agents and tables, for a team to assemble on its own data, with seats priced in the dozens. Kamu is one worker for one product. It decides what to do from a plan it wrote, does it inside hard caps, and asks you before it spends money. If you have nobody to build workflows, Kamu is the alternative. If you do, Copy.ai's ceiling is far higher.

Q: Can I build my own workflows in Kamu?
A: No. You choose which jobs are on: finding customers, writing articles, posting on social, a landing page, directory listings and Meta ads. You set the pace and approve anything that spends money or changes the plan. How each job is done is Kamu's decision, logged with the reason, and you steer it in chat rather than in a builder. That is the trade: there is nothing to set up, and also nothing to tune.

Q: Does Kamu connect to my CRM?
A: No. Kamu keeps its own record of the people it found, what it wrote to them and what they replied, and shows you that in the app and in the morning standup. Copy.ai's Tables and its CRM enrichment workflows are built for teams whose data already lives in a CRM and needs to stay there. If that is you, Kamu will feel like a silo.

Q: Which is cheaper for one person?
A: Kamu, by a margin. Copy.ai's Chat plan is $29 a month, or $24 billed annually, as of September 2026, and it includes five seats one person will not use; workflows begin on Growth at $1,000 a month billed annually. Kamu's Seed plan is €25 a month for one product, with daily caps of 10 emails and 2 X posts, and Starter is €49. The two are hard to compare, because one buys a platform and the other buys the work.

## Reply.io alternatives for a founder doing outbound alone

https://kamugrows.com/compare/reply-io

Reply.io is a sales engagement platform with multichannel sequences, a contact database and an AI SDR agent. Claims checked 2026-09-03.

Verdict: Reply.io is the better product if a sales team is going to run sequences across email, LinkedIn and calls and wants the controls for that. Kamu is the better product if nobody on your side will run anything: it writes a few emails a day on its own, answers the replies, and covers content, social, a landing page, listings and ads as well.

Reply.io calls itself an AI sales outreach and cold email platform, and the description is fair. You build conditional sequences that step through email, LinkedIn touches, WhatsApp, SMS and calls, with anything else reachable through Zapier. Contacts come from a database it says holds over a billion records. HubSpot, Salesforce and Calendly plug in so a booked meeting lands in the pipeline. There is also Jason, an AI SDR agent sold as its own plan under the line "Sales on autopilot". As of September 2026 the pricing page lists the email plan from $59 a month and the multichannel plan from $99, then the AI SDR plan from $500 and an agency plan from $166, with LinkedIn automation at $69 per account and calls and SMS at $29 per account on top. The trial is 14 days.

Everything about that is designed around a sales team. Plans are metered in active contacts, the number of people you can open a sequence on each month, which is the right unit when a rep's job is to work a list. Add-ons are per account because there are several accounts. Reporting is for a manager who wants to know which step converts.

Kamu is built for the case where there is no team and no list. You give it a product URL. It reads the site, works out who buys something like it, finds those people, checks each address before writing, and sends between 10 and 200 emails a day depending on plan, one at a time, each opening on something that person shipped or wrote. When someone replies it reads the thread and answers on its next run. Then it goes and does the other five jobs: articles for phrases with buyers behind them, posts to X, LinkedIn and Instagram from what actually shipped, a landing page it measures and rewrites, directory listings, and a video ad built paused.

The overlap is the cold email. The rest of each product is aimed at different people.

- What it is. Kamu: A growth agent that does the outbound itself, then the content, social, landing page, listings and ads. Reply.io: A sales engagement platform: sequences, a contact database, deliverability tooling and an AI SDR agent, operated from a seat.
- Who does the work. Kamu: Kamu. It picks the people, writes each email and answers the replies, unattended. Reply.io: Your team, from the sequence builder. The AI SDR plan hands part of that to Jason.
- Channels. Kamu: Email from your own domain, and LinkedIn messages when you connect an account. It makes no calls and sends no SMS. Reply.io: Email, LinkedIn, WhatsApp, SMS and calls in one sequence, plus anything reachable through Zapier.
- Volume. Kamu: 10 emails a day on Seed, 25 on Starter, 75 on Growth, 200 on Scale. Reply.io wins this row. Reply.io: Metered by active contacts, 1,000 a month on the email plan and 5,000 on multichannel, with unlimited emails and follow-ups to each.
- What it costs. Kamu: €25, €49, €149 or €349 a month, published, everything included. Reply.io: From $59 a month for email, $99 multichannel, $500 for the AI SDR plan, as of September 2026, plus $69 per LinkedIn account and $29 per account for calls and SMS.
- Where the names come from. Kamu: Kamu reads your product and goes looking for people who match, then checks each address before it writes. Reply.io: A database of over a billion contacts, by Reply.io's count, searched from inside the platform on data credits.
- What is refused. Kamu: An unverified address, a second email after a no, any send to a suppressed address across every product in your account, and open-tracking pixels. Reply.io: Reply.io describes a deliverability suite. Whether a step goes out is decided by the sequence you built.
- What else it does. Kamu: SEO articles published through Ghost or WordPress after approval, social posts from what shipped, a landing page, directory listings and a paused Meta video ad. Reply.io: Outreach and the tooling around it, including meeting booking and CRM sync. It is not trying to write your blog.

When Reply.io is the better choice: Take Reply.io if you have reps. A sales engagement platform earns its price when several people are working sequences at once and a manager needs to see which step and which channel is converting, and Reply.io shows that in a way an agent writing one email at a time never will. Channels are a clean win: if the plan is email plus a LinkedIn touch plus a call, Kamu makes no calls and has no SMS or WhatsApp. Volume is another, since an active-contact meter with unlimited follow-ups is built for thousands of first touches a month and Kamu stops at 200 emails a day. So is control, because you decide every word of every step and can change it this afternoon. And if you already keep a CRM, HubSpot and Salesforce sync and Calendly booking are on the page, and a team that lives in those tools should buy the thing that talks to them.

Q: Is Kamu a Reply.io alternative or a different kind of product?
A: A different kind, with one overlap. Reply.io is software a sales team operates to run outreach across several channels. Kamu is a worker that runs outreach on its own and also does content, social, a landing page, listings and ads. If the question is which one sends your sequences, Reply.io does, and Kamu has no sequences at all. If the question is who does the outbound when you have no time for it, that is Kamu's job.

Q: Does Reply.io's Jason do the same thing as Kamu?
A: They overlap on outreach. Jason is an AI SDR agent inside Reply.io, sold on a plan listed from $500 a month as of September 2026, and it works the outbound within the platform's sequences and data. Kamu's outreach is one of six objectives and the whole product is priced from €25 a month. Which is better depends on whether you want an agent inside a sales platform your team already uses, or one agent that covers growth for a company with no sales team.

Q: Can Kamu send LinkedIn messages and make calls like Reply.io?
A: LinkedIn, yes, once you connect an account: Kamu sends a small number of messages a day within your plan's cap. Calls, SMS and WhatsApp, no. Reply.io runs all of those in one sequence, and if that is the motion you want it is the better buy by some distance.

Q: Which is cheaper for one founder?
A: Kamu, per month, at every plan. Reply.io's email plan is listed from $59 a month with unlimited emails to 1,000 active contacts, so per email sent it is far cheaper than Kamu, which sends 10 to 200 a day. The question underneath is whether you have a list to send to and a person to run the sequence. If you do, Reply.io's price buys more. If you have neither, the cheaper platform still produces nothing.

## A Snov.io alternative that picks the people and writes the email

https://kamugrows.com/compare/snov-io

Snov.io is an email finder and verifier with drip campaigns, a CRM and LinkedIn automation, run from your own seat. Claims checked 2026-09-03.

Verdict: Snov.io is the cheaper and better buy for a person who wants to find addresses, verify them and run drip campaigns themselves. Kamu finds and verifies too, and then does the part Snov.io leaves to you: deciding who is worth writing to, writing each email, and answering what comes back.

Snov.io is a toolbox for outbound. The headline on its site is find leads, enrich them with emails and phone numbers, send cold emails, and the product list underneath is long: an email finder and a bulk version, a LinkedIn finder and a browser extension, a verifier, mailbox warm-up, a deliverability test, drip campaigns, an AI email writer, an email tracker, a small sales CRM, and LinkedIn automation as a paid slot. As of September 2026 the free trial gives 50 credits and 100 recipients with no card, Starter is $39 a month for 1,000 credits and 5,000 recipients with unlimited emails and unlimited team seats, Pro S is $99 for 5,000 credits and 25,000 recipients, and the ladder runs on through Pro M at $189 and Pro L at $369 to Ultra at $738. LinkedIn automation is $69 a month per slot on top.

Every one of those tools waits for you. You decide who the buyer is and search or upload the list. Somebody writes the sequence and sets the delays. Replies land in your own mailbox. Snov.io makes each step quicker and cheaper than doing it by hand, which is what a good tool does.

Kamu does the steps. It reads your product and works out who buys something like it, then goes looking for those people and checks each address before it writes, so a bounce is the exception rather than the report. It writes one email at a time, opening on something that person shipped or wrote, inside a daily cap of 10 to 200 depending on plan. When someone replies it reads the thread and answers on its next run. A no closes the thread, and an unsubscribe suppresses the address across every product in your account. Then it moves on to articles, social posts, a landing page, directory listings and a video ad, because outreach is one of six things it does.

Snov.io is a very good toolbox at a low price. Kamu is a worker, priced from €25 a month and rising with how much it does.

- What it is. Kamu: A growth agent that finds people, writes to them and answers, and also does content, social, a landing page, listings and ads. Snov.io: A set of outbound tools: email finder, verifier, warm-up, drip campaigns, a CRM and LinkedIn automation.
- Who decides the target. Kamu: Kamu, from reading your product and your plan. It learns who says yes and looks for more of them. Snov.io: You. Snov.io searches by the filters you give it or works from a list you upload.
- Who writes the email. Kamu: Kamu, one at a time, each opening on something that person shipped or wrote. Snov.io: You, in the drip campaign editor. An AI email writer will draft one when asked.
- Who answers the reply. Kamu: Kamu reads the whole thread and writes back on its next run. A no closes it. Snov.io: You. Replies land in your own mailbox.
- Verification. Kamu: Every address is checked before anything is sent, and delivery is recorded for every send. Snov.io: A verifier you run, in bulk or one at a time, with an accuracy figure of 98% by Snov.io's own count.
- What it costs. Kamu: €25 a month on Seed, €49 Starter, €149 Growth, €349 Scale. Snov.io: A free trial with 50 credits; Starter $39 a month, Pro S $99, up to Ultra at $738, as of September 2026. LinkedIn automation is $69 per slot.
- Volume. Kamu: 10 to 200 emails a day by plan. Snov.io wins this row by a distance. Snov.io: Unlimited emails on every paid plan, with recipients metered from 5,000 a month on Starter.
- What else it does. Kamu: SEO articles through Ghost or WordPress after approval, social posts from what shipped, a landing page, directory listings, a paused Meta video ad. Snov.io: Outbound and the CRM around it. It does not write your blog or run your socials.

When Snov.io is the better choice: Pick Snov.io if you will run it. A founder who knows exactly who to write to, has an hour a week for it and wants the cheapest reliable way to find addresses, verify them and send a sequence gets more per dollar from Snov.io than from anything on this site. The Starter plan's 5,000 recipients a month is more outreach than Kamu sends on any plan, and the free trial means you can check the finder against your own list before paying. Snov.io also covers ground Kamu does not: phone numbers, a browser extension for pulling addresses off LinkedIn, a small CRM, and LinkedIn automation as an add-on. Agencies and sales teams get unlimited seats on every plan, which is the right shape for several people sharing one toolbox. If your problem is the mechanics of outbound rather than the time to do it, buy the toolbox.

Q: Does Kamu find and verify email addresses like Snov.io?
A: Yes, and it does it before it writes rather than as a separate job. Kamu looks for people who match your buyers, finds a working address for each, checks it, and records delivery on every send. What it does not offer is a standalone finder or verifier you can point at your own list, a browser extension, or phone numbers. If you want those tools in your hands, Snov.io has them.

Q: Is Kamu cheaper than Snov.io?
A: Kamu's Seed plan is €25 a month and Snov.io's Starter is $39, so the entry price is close. Per email the comparison is not close: Snov.io sends unlimited emails to 5,000 recipients a month on Starter, and Kamu sends 10 a day on Seed and 200 a day on Scale. Kamu's price buys the deciding and the writing as well as the sending, and buys the other five jobs. Snov.io's buys tools.

Q: Can I use Snov.io and Kamu together?
A: Yes, and the split is clean. Snov.io is useful for the list you already have and want to work yourself, or for one-off lookups. Kamu is for the outreach nobody on your side has time to run, plus the content and social. They do not write to the same people for the same reason, so there is nothing to coordinate.

Q: Does Snov.io have an AI agent that runs outreach for you?
A: Its site lists an AI email writer, which drafts a message when you ask, and drip campaigns with LinkedIn automation as a paid slot. Those are tools you operate. Nothing on the site describes an agent that decides who to contact and runs the conversation on its own, which is the part Kamu does, and it is the reason the two are not really competing for the same buyer.

## An Arcads alternative that builds the campaign too

https://kamugrows.com/compare/arcads

Arcads is an AI video ad generator with a library of AI actors, for performance marketers who run the ads elsewhere. Claims checked 2026-09-03.

Verdict: Arcads is the better product for a marketer who writes scripts, wants many actors and formats to choose from, and runs the ads somewhere else. Kamu makes one short presenter video from what your product does, refuses any line where the presenter claims to be a customer, and builds the Meta campaign around it, paused.

Arcads makes video ads with AI actors. You write the script, pick from a library its site puts at over a thousand actors or build an avatar of your own, and it renders a person delivering your lines to camera in the style of user-generated content. It translates into more than 30 languages, sets the emotion from how you write the line, adds B-roll, music and captions, and can put your product in the actor's hands. Its customers, in its own words, are fast growing D2C brands, mobile app studios and marketing agencies, and the SaaS page describes an infinite canvas for creating, testing and scaling ads as a team. Pricing is not published on the site as of September 2026. There is a subscription with credits, and the help centre covers buying more of them.

What Arcads does not do is the rest of the ad. It is a creative tool. The media plan, targeting, budget and the account it runs in are yours, and the assumption is that a performance marketer is on the other side producing variants and testing them.

Kamu starts from the product rather than a script you wrote. It reads your site, writes a script around what the product does, casts one of six generated presenters, and renders a short vertical video with captions. It then writes the ad copy, picks the countries, sets a daily budget, and builds the campaign, ad set and ad in your Meta account, all paused, with a ceiling on the budget checked in code. Without an ad account connected it drafts the whole thing and leaves it for you to run by hand. Once an ad has been seen a thousand times it reads back spend, clicks and cost per click, and it will pause one that spends without working. It never starts one.

One rule sits under all of it: the presenter is never a customer. They can say what the product does. A script that has them claim to use it, or to have got a result from it, is refused before anything renders.

- What it makes. Kamu: One short vertical video: a script from your product, a generated presenter, voice and captions. Arcads: AI actor videos from your script, with a library of over a thousand actors, custom avatars and translation into more than 30 languages.
- Who writes the script. Kamu: Kamu, from what the product does, with no invented numbers and no fabricated testimonials. Arcads: You. Arcads renders what you wrote, and you set the emotion by how you write it.
- Who runs the ad. Kamu: Kamu builds the campaign, ad set and ad in your Meta account, paused. Only you can start it. Arcads: You, in whatever ad platform you use. The site describes creating, testing and scaling ads; placing them is your side of the work.
- Variety. Kamu: A cast of six presenters and rooms taking turns. One ad at a time, built when there is nothing to run. Arcads: Many actors, formats and languages, built for producing variants to test. Arcads wins this row.
- What is refused. Kamu: Any line where the presenter claims to use the product or to have got a result from it. A campaign that would start running. Spend above the daily ceiling. Arcads: Not published as a list on the site.
- What it costs. Kamu: Part of a plan from €25 a month. A twenty-second clip costs about 100 credits, roughly double the first few times while the face is invented. Ad spend sits on your own Meta account. Arcads: Not published on the site as of September 2026. A subscription with credits; buying more is covered in the help centre.
- Without an ad account. Kamu: It still makes the video and drafts the whole ad, copy, targeting and budget, for you to run by hand. Arcads: Arcads never needed one. The video is the product, and where it runs is your decision.
- After it runs. Kamu: Spend, impressions, clicks and cost per click, read back once a thousand people have seen it. An ad spending without clicks is paused. Arcads: Whatever your ad platform reports. Arcads is not in that loop.

When Arcads is the better choice: Buy Arcads if you are producing ad creative in volume. A performance marketer testing ten hooks against three audiences a week needs many faces, many formats and a fast way to swap a line and re-render, and a library of over a thousand actors with translation into thirty-odd languages is exactly that. Kamu's cast of six and its one-ad-at-a-time pace is a bottleneck for that person. Arcads also wins wherever the ads run outside Meta: on TikTok or YouTube Kamu has nothing to offer, since Meta is the only ad account it builds in. And it wins for anyone who wants an actor to pick up the product and demo it, to speak in a customer's voice, or to follow a script written by a human who knows the audience, because Kamu's presenter rule refuses the customer voice on purpose and its scripts come from the product page rather than from you. If a team on your side already runs paid, give them Arcads.

Q: Does Kamu make UGC-style video ads like Arcads?
A: It makes a short vertical video of a generated presenter talking to camera, with captions, which looks like the same genre. The difference is what the presenter is allowed to say. Arcads renders your script, including a customer's voice if you write one. Kamu's presenter can explain what the product does and cannot claim to use it, and a script that tries is refused before rendering.

Q: Can Kamu use the videos I made in Arcads?
A: No. Kamu builds campaigns only around videos it rendered itself, and there is no way to hand it a file. If you have Arcads creative you want to run, build that campaign in Ads Manager yourself and let Kamu take the parts of growth that have nothing to do with ads: the outreach, articles, social posts and listings.

Q: Which platforms can Kamu run ads on?
A: Meta only, and only paused. Connect an ad account and Kamu creates the campaign, ad set and ad there with a daily budget ceiling checked in code, and you start it in Ads Manager. There is no TikTok, YouTube or Google Ads support. Arcads makes creative you can take to any of those, which is a real advantage if Meta is not where your buyers are.

Q: Why will Kamu not let the presenter say they use the product?
A: Because they do not, and an ad that says so is a fabricated testimonial. Kamu holds the same line everywhere it writes: it does not invent numbers or ratings, and it does not put words in the mouth of a customer who does not exist. The presenter can describe the product plainly and name it once. That is a narrower ad than Arcads will render, and it is narrower on purpose.

## Hiring a SaaS marketing agency, seen from the buyer's side

https://kamugrows.com/compare/saas-marketing-agency

A SaaS marketing agency is an outside firm that runs some or all of your marketing on a monthly retainer. Claims checked 2026-09-03.

Verdict: A SaaS marketing agency brings people who have grown software companies before, on a retainer with a term, an onboarding month and a reporting rhythm. Kamu brings a worker that starts today for a plan price and shows every action with its reason. Which is right depends on whether you are buying experience or hours.

Here is what you are agreeing to when you sign with a SaaS marketing agency, in the order it happens.

A discovery call, then a proposal. The proposal describes a retainer: a fixed monthly fee for a defined scope, usually with a minimum term and a notice period. The first month is onboarding. Kalungi, one of the agencies on the first page for this search, describes its first month as installing the foundation: ICP and messaging, positioning, a 30/60/90/180 plan, website and brand alignment, and a reporting baseline. After that you settle into a cadence, a weekly or fortnightly call and a monthly report, and Kalungi's own estimate of the client's time is two to five hours a week on reviews and syncs.

Agencies specialise. Some do paid media and nothing else; others do SEO and content, or brand and design; a few run the whole function. Kalungi's full-service tier is a useful reference for the top of the market: it says it replaces ten or more hires with one accountable team, puts the cost at $50,000 and up a month, and aims it at SaaS between $5M and $50M in revenue. Smaller agencies charge less, and almost nobody publishes a rate.

What you are buying is people who have done this before. An agency has run a launch, rebuilt a category page until it ranked, spent a real paid budget and learnt from it. That transfers, and it is the honest reason to pay a retainer.

Kamu is a different purchase. It is a plan from €25 to €349 a month with no term and no onboarding call. Give it a URL and it starts: cold email to people who match your buyer, articles for phrases with buyers behind them, posts from what shipped, a landing page, listings, a paused video ad. It has never done this before in any way that transfers, and it will not tell you your strategy is wrong. It will show you every action it took and why, the same morning.

- What you sign. Kamu: A monthly plan, €25 to €349, cancel any month. A SaaS marketing agency: A retainer for a scope, usually with a minimum term and a notice period.
- The first month. Kamu: Kamu reads your site, writes a plan and starts the same day. A SaaS marketing agency: Onboarding. Kalungi describes its first month as ICP, messaging, positioning, a 30/60/90/180 plan and a reporting baseline.
- Who does the work. Kamu: One agent, every day, inside the plan's daily caps. There is no account manager. A SaaS marketing agency: A team at the seniority the retainer buys. At Kalungi, an Associate CMO backed by an execution team.
- Your time. Kamu: A short standup each morning and the approvals that spend money or change the plan, in chat, Slack, Telegram or WhatsApp. A SaaS marketing agency: Calls and reviews. Kalungi puts it at two to five hours a week for its full-service clients.
- What it costs. Kamu: €25 on Seed, €49 Starter, €149 Growth, €349 Scale, published. A SaaS marketing agency: Rarely published. Kalungi's full-service tier is $50,000 and up a month as of September 2026; smaller agencies charge less and quote per scope.
- Experience. Kamu: None that transfers. It reads your market fresh and learns from what your buyers reply. A SaaS marketing agency: The reason to pay: people who have launched, ranked and spent budget before, for companies like yours.
- What you see. Kamu: Every email, post, article, listing and ad, with the reason, as it happens. A SaaS marketing agency: A report on the cadence you agreed, written by the agency, plus whatever the calls surface.
- Getting out. Kamu: Cancel any month. Prospects, drafts, articles and the landing page stay in your account. A SaaS marketing agency: Notice, then a handover: ad accounts, analytics, content and the passwords to all of them.

When A SaaS marketing agency is the better choice: Hire the agency when what you need is judgement earned elsewhere. A launch that has to land on a date, a paid budget in the tens of thousands, a category you are new to, a brand that needs to be designed rather than described: these are jobs where somebody who has done it three times before saves you the expensive version of learning. An agency also brings breadth Kamu cannot: design, PR, video shot with a camera, partnerships, events, and a senior person who will say your positioning is wrong. If you are past a few million in revenue and marketing is now a function rather than a task, an agency's structure fits, and the retainer is small against what a wrong quarter costs. And if you want a supplier who owns the outcome, with a contract and a person to ring when it is not working, that is what a retainer buys and what a plan price never will.

Q: What does a SaaS marketing agency cost?
A: Few publish a price. The one figure on this page comes from Kalungi, which lists its full-service engagement at $50,000 and up a month for SaaS between $5M and $50M in revenue as of September 2026, and aims cheaper tiers at earlier companies. Smaller agencies charge less and quote per scope. Kamu is €25 to €349 a month, published, and the same for everyone.

Q: How long does an agency take to show results?
A: Longer than the first invoice. Kalungi says its clients see clarity in month one, movement in month two and qualified opportunities by month three, and that is the honest shape of a retainer: onboarding, then work, then pipeline. Kamu starts the day you add a URL, which is not the same as saying it produces pipeline that day. The difference is that you can watch the work from the first morning.

Q: Is Kamu a SaaS marketing agency?
A: No. An agency is a company of people you retain; Kamu is software that runs six growth jobs on its own and reports to you. It has no account manager and no creative team. It costs a plan price, starts the same day, and logs everything it does with the reason. If the word agency is doing the work of meaning somebody else handles marketing, Kamu qualifies. If it means experienced people, it does not.

Q: When should a founder pick the agency over Kamu?
A: When the job needs experience you cannot get from a log. A launch on a date, a paid budget large enough that a wrong week hurts, a rebrand, a category you have never sold into. Also when you are big enough to get the senior people's attention, since the seniority you get tends to track the size of the retainer, and a small account at a large agency is a different product from the one that was pitched.

## An AI SDR software alternative for a founder with no sales team

https://kamugrows.com/compare/ai-sdr

An AI SDR is a category of software that runs sales development, prospecting and outreach for sales teams. Claims checked 2026-09-03.

Verdict: AI SDR software runs outbound at volume for a sales team: hundreds or thousands of researched contacts a month, sequences across email and LinkedIn, meetings booked into a CRM, priced by contact allowance or by quote. Kamu sends a few good emails a day and then does the rest of growth, for a founder who has no sales team to hand the meetings to.

An AI SDR agent does the job a sales development rep does: builds a list, researches each person, writes and sends the outreach, works the replies and books the meeting. Three vendors show what the category looks like from the buyer's side.

AiSDR sells self-serve. As of September 2026 its Solo plan is $250 a month for 200 AI-researched contacts, one user, one domain and three mailboxes, month to month. Explore is $900 a month for 800 contacts, billed quarterly, and Scale is $2,500 a month for 2,500 contacts with Salesforce sync. It sends by email and LinkedIn, syncs with HubSpot, and names revenue teams, sales leaders and founders as its buyers.

11x sells Alice, an outbound digital worker across email, phone, LinkedIn and SMS, and Julian, an inbound phone agent. There is no price on the site; you book a demo. It names sales, RevOps and marketing teams at mid-market and enterprise companies.

Artisan sells Ava, an autonomous AI BDR that finds and enriches leads, writes and sends, handles replies and books meetings, with data, deliverability, sequencing and a dialer in one platform. The homepage routes to a demo request.

The shape is consistent. AI SDR tools are metered in researched contacts or quoted per company, they assume a CRM to write into and someone to take the call, and the more capable tiers are sold through a conversation. That is the right design for a team with a quota.

Kamu is designed for the founder who has none of that. It writes between 10 and 200 emails a day depending on plan, one at a time, each to a person it found and verified and each opening on something that person shipped or wrote. It answers replies on its next run and stops when someone says no. There is no per-seat price because there is one of you. Outreach is one of six jobs: the same agent writes articles, posts from what shipped, keeps a landing page, files listings and builds a paused video ad. The plans are €25 to €349 a month, published.

- What it is. Kamu: A growth agent: outreach plus content, social, a landing page, listings and ads, for one founder. An AI SDR: Software that runs sales development for a team: prospecting, sequences, replies and booking, into a CRM.
- Who it is for. Kamu: A founder with no sales team, who will take the reply themselves. An AI SDR: Sales and revenue teams with quota. AiSDR names revenue teams and sales leaders; 11x names sales, RevOps and marketing at mid-market and enterprise.
- Volume. Kamu: 10 to 200 emails a day by plan, one at a time. The category wins this row. An AI SDR: AiSDR meters 200, 800 or 2,500 researched contacts a month by plan; 11x and Artisan do not publish limits.
- How you buy. Kamu: Self-serve, monthly, cancel any month. €25, €49, €149 or €349. An AI SDR: AiSDR from $250 a month self-serve, with quarterly billing above that; 11x and Artisan by demo and quote, as of September 2026.
- Channels. Kamu: Email from your own domain, and LinkedIn messages when connected. No phone. An AI SDR: Email and LinkedIn everywhere; 11x adds phone and SMS, and Artisan includes a dialer.
- Where the meetings go. Kamu: Into your calendar. Kamu drops your meeting link into the thread when someone asks for a call. An AI SDR: Into the CRM: HubSpot and Salesforce sync at AiSDR, CRM integration at 11x, and a team to take the call.
- What is refused. Kamu: An unverified address, a follow-up after a no, any send to a suppressed address across every product you run, and open-tracking pixels. An AI SDR: Varies by vendor and is rarely published as a list. Check each one's unsubscribe and deliverability handling before you sign.
- What else it does. Kamu: Articles through Ghost or WordPress after approval, posts from what shipped, a landing page, directory listings and a paused Meta video ad. An AI SDR: Sales development, and the data and dialer around it. Marketing is somebody else's product.

When An AI SDR is the better choice: Buy AI SDR software when you have a sales team. If reps carry quota, a CRM is kept clean and there are people to take meetings the moment they land, an agent metered in hundreds or thousands of researched contacts a month is the right instrument, and Kamu's cap of 200 emails a day on its largest plan is a ceiling you would hit in the first week. The category also wins on the machinery of sales: two-way CRM sync, a dialer, phone and SMS in the sequence, and at 11x and Artisan a sales-led onboarding with a named person accountable for making it work. If outbound is your only gap and a specialist beats something broader, AiSDR's Solo plan at $250 a month is a real self-serve option for a small team. And if what you want to buy is meetings, in volume, with a manager watching the funnel, none of that is what Kamu is for.

Q: Is Kamu an AI SDR agent?
A: For one of its six jobs, yes. Kamu finds people, writes to them one at a time, answers replies and stops on a no, which is the sales development loop. It is not metered in contacts or seats and it does not write into a CRM. The same agent then goes and does the rest of growth. If you want only the SDR part at volume, a product built for a sales team goes deeper on it.

Q: What does AI SDR software cost?
A: It depends on who publishes. AiSDR lists Solo at $250 a month, Explore at $900 and Scale at $2,500 as of September 2026, metered by researched contacts. 11x and Artisan do not publish prices and sell through a demo. Kamu is €25 to €349 a month with outreach included in every plan. Per contact reached the AI SDR tools cost less. Per month, for one founder, Kamu does.

Q: Do I need a CRM to use Kamu?
A: No. Every person Kamu writes to has one row in its own leads list, with everything both sides said, and your meeting link goes into the thread when somebody asks for a call. AI SDR software assumes a CRM and a team behind it, which is why HubSpot and Salesforce sync sit on the front page at AiSDR and 11x. If you already run one, that is a point in the category's favour.

Q: Which AI SDR tools does this site compare directly?
A: Artisan and 11x each have their own page, and Soleur, another autonomous agent, has one too. The sending platforms Instantly, Smartlead, lemlist and Reply.io are compared as tools you operate rather than as agents. Every figure on those pages was read from the vendor's own site on the date shown, and where a vendor publishes no price the page says so.

# Glossary

## Cold email

https://kamugrows.com/glossary/cold-email

Cold email is a first email to someone who has not asked to hear from you, sent because something about their situation suggests they would want what you sell.

Cold email is a first email to someone who has not asked to hear from you. It is not defined by being unwanted or by being sent in bulk. It is defined by the absence of a prior relationship, which is why the same message can be legitimate business correspondence or spam depending on facts that have nothing to do with its wording.

Three things separate one from the other. The first is relevance: a real reason this person, specifically, would care, which usually means something they built, wrote, hired for, or announced. The second is a lawful basis to contact them at all, which in the EU and UK means a legitimate interest you can articulate, and in the US means honouring the CAN-SPAM rules on identification and opt-out. The third is a working way out, honoured immediately and without conditions.

The common mistake is treating volume as the lever. Sending more of a message that does not earn a reply lowers the reply rate, raises the complaint rate, and damages the domain doing the sending, so the same campaign gets worse over time rather than better. The number that matters is replies per thousand sent, and it moves on relevance rather than on quantity.

In practice:
- An email to a founder who just posted about hiring their first salesperson, from a company that sells sales tooling.
- A note to an engineering lead referencing an open-source library they maintain, from a company whose product depends on it.
- Not cold email: a follow-up to somebody who downloaded a whitepaper and gave you their address.

Q: Is cold email legal?
A: In most jurisdictions yes, business-to-business, with conditions. CAN-SPAM in the US requires accurate headers, a clear identification of who is sending, a physical postal address, and an opt-out you honour within ten business days. It does not require prior consent. GDPR and PECR in the EU and UK are stricter: you need a lawful basis, usually legitimate interest, which means being able to explain why this specific recipient would reasonably expect to hear from you, and you must tell them where their data came from. Consumer email is a different matter and generally needs consent. None of this is legal advice, and the rules for your country and industry are worth checking properly.

Q: How is cold email different from spam?
A: Spam is unsolicited bulk email sent without regard to whether the recipient is a plausible buyer, usually with a disguised or missing sender and no honest way out. Cold email is unsolicited but not bulk in that sense: it is aimed, it identifies the sender truthfully, and it takes one click to stop. The recipient is the one who decides which of the two you sent, and the complaint rate on your domain is how you find out.

Q: How many cold emails a day is safe to send?
A: It depends on the age and reputation of the sending domain rather than on any universal number. A new domain sending at volume is the single clearest spam signal there is, which is why sending ramps over weeks rather than starting at full rate. A mature domain with a good reputation and a low complaint rate can sustain far more than a new one. The right cap is the one where your bounce and complaint rates stay flat as volume rises; when they climb, you are already sending too much.

## Email deliverability

https://kamugrows.com/glossary/email-deliverability

Deliverability is whether an email you send arrives in the inbox rather than the spam folder or nowhere at all. It is decided by the receiving provider, not by you.

Deliverability is whether the email you send arrives in the inbox. It is distinct from delivery, which only means the receiving server accepted the message: an email can be delivered and still land in spam, where it counts as delivered by every sending tool and is read by nobody.

The decision belongs to the receiving provider, and it is made on evidence about the sender rather than on the content of the message. The inputs are the sending domain's history, whether the message is authenticated, how people who received your last emails behaved, and how closely this send resembles patterns the provider associates with bulk mail. Content matters far less than most senders assume, which is why rewriting a subject line rarely fixes a deliverability problem.

What is actually in your control is narrow but sufficient: authenticate properly with SPF, DKIM and DMARC; verify addresses before sending so bounces stay low; ramp a new domain slowly rather than starting at volume; make unsubscribing trivial so people use it instead of the spam button; and stop sending to anyone who has not engaged. All five are about the sender rather than the copy.

Also called: Deliverability.

In practice:
- A message accepted by Google and filed straight into spam: delivered, not deliverable.
- A new domain sending five hundred emails on its first day, with almost all of them filtered.
- A domain whose complaint rate crosses roughly 0.3%, after which Gmail begins filtering aggressively.

Q: Why do my emails go to spam even though nobody reported them?
A: Because filtering is predictive rather than punitive. A receiving provider decides before anyone has read the message, using the sending domain's history, whether the mail is authenticated, and how similar the send looks to bulk patterns it has filtered before. A brand new domain with no history is treated as suspicious by default, and an unauthenticated message from one is treated as very suspicious. Absence of complaints is not evidence of good standing; it is usually the absence of a track record.

Q: Does a spam-word checker help?
A: Very little. The idea that particular words trigger filters is largely folklore left over from filters of twenty years ago. Modern filtering weighs sender reputation, authentication and recipient behaviour far above wording. Time spent removing the word "free" is better spent verifying your list and fixing your DMARC record.

Q: What is a good bounce rate?
A: Under 2%, and under 1% if you can manage it. Bounces are the fastest way to damage a sending domain because they demonstrate you are mailing addresses you have not checked, which is what a list bought rather than built looks like from the receiving side. Verifying addresses before sending is the fix, and it is cheap relative to the damage.

## Email authentication

https://kamugrows.com/glossary/email-authentication

Email authentication is a set of three DNS records, SPF, DKIM and DMARC, that together let a receiving server verify an email genuinely came from the domain it claims.

Email authentication is how a receiving server checks that a message claiming to be from your domain really is. It rests on three DNS records that do different jobs, and it works only when all three are present.

SPF, the Sender Policy Framework, is a list of the servers allowed to send mail for your domain. The receiver compares the sending server against that list. It breaks on forwarding, because a forwarded message arrives from a server your record never named.

DKIM, DomainKeys Identified Mail, attaches a cryptographic signature to each message using a private key, which the receiver verifies against a public key in your DNS. It proves the message was authorised by the domain and was not altered on the way, and unlike SPF it survives forwarding.

DMARC ties the two together and is the part most senders skip. It tells receivers what to do when a message fails both checks, and it requires alignment: the domain in the visible From address has to match the domain that passed SPF or DKIM. Without DMARC, an attacker can pass SPF for their own domain while showing your name in the From field. Since 2024, Google and Yahoo require DMARC on any domain sending meaningful volume, so it is no longer optional in practice.

Also called: SPF, DKIM, DMARC.

In practice:
- An SPF record: v=spf1 include:_spf.google.com ~all
- A DMARC record starting at monitoring only: v=DMARC1; p=none; rua=mailto:dmarc@example.com
- The same record after a month of clean reports: v=DMARC1; p=reject

Q: Do I need all three, or is SPF enough?
A: All three. SPF alone fails on any forwarded message and does nothing about the From address a recipient actually reads, which is the field that gets spoofed. DKIM alone proves authorisation but tells receivers nothing about what to do with failures. DMARC is what makes the other two enforceable, and it is what Google and Yahoo now check on domains sending at volume.

Q: Should I set DMARC to reject straight away?
A: No. Start at p=none, which asks receivers to report failures without acting on them, and read the reports for a few weeks. Almost every organisation discovers a legitimate sender it had forgotten about, such as a billing system or a help desk, that is not covered by its SPF or DKIM setup. Moving to p=reject before finding those means silently destroying your own mail.

Q: Does authentication get me into the inbox?
A: No, it gets you considered. Authentication is a precondition rather than an advantage: failing it is close to fatal, and passing it earns you nothing except the chance to be judged on your sending history and how recipients respond. It is table stakes, not a lever.

## Domain warmup

https://kamugrows.com/glossary/domain-warmup

Domain warmup is raising the volume sent from a new domain gradually over weeks, so receiving providers build a record of it behaving well before it sends at full rate.

Domain warmup is the practice of raising sending volume from a new domain slowly, over weeks, instead of starting at the rate you eventually want. Email warmup is the same practice under a more common name, and what gets warmed is the domain and the IP the mail leaves from rather than any one mailbox. It exists because a receiving provider has no history for a new domain and treats absence of history as risk. A domain that has never sent anything and then sends a thousand emails in a day looks exactly like a domain bought that morning for spam, because that is very often what it is.

A warmup starts at a handful of messages a day and roughly doubles each few days, with the ramp paused whenever bounce or complaint rates move. What is being built is not a volume allowance but a record: authenticated mail, sent to addresses that exist, to people who open and reply and do not complain. Engagement is the part that matters most, which is why warming up by sending to addresses that will never reply achieves very little.

Two practical notes. Warmup applies per sending domain and per IP, so moving to a new domain or a new provider means starting again. And a subdomain inherits some but not all of the parent domain's standing, which is why sending outreach from a subdomain is a common way to protect the reputation of the domain your actual business email runs on.

Also called: Email warmup, IP warmup.

In practice:
- Twenty emails a day in week one, forty in week two, eighty in week three, pausing whenever complaints rise.
- Sending outreach from mail.example.com rather than example.com, so a bad campaign cannot damage the domain your invoices come from.
- Starting over from a low rate after switching sending providers, because the new IP has no history.

Q: How long does warming up a domain take?
A: Three to six weeks to reach a normal outreach volume, longer if you want high volume. The ramp is bounded by how fast providers accumulate evidence rather than by anything you control, so the process cannot be shortened by sending more. It can easily be lengthened by sending too much and having to back off.

Q: Do warmup services that exchange opens actually work?
A: Partly, and less than they claim. Networks where seed accounts open and reply to each other's mail do generate engagement signals, but the engagement is not from real prospects and providers have got better at recognising the pattern. It builds some history; it does not substitute for sending relevant mail that real people answer.

Q: Should I use a separate domain for outreach?
A: Usually yes. Outreach carries a risk of complaints that ordinary business mail does not, and a damaged domain affects everything sent from it, including invoices, password resets and replies to customers. A separate domain or subdomain contains that risk. The trade-off is that it starts with no reputation and has to be warmed.

## Sender reputation

https://kamugrows.com/glossary/sender-reputation

Sender reputation is the standing a mailbox provider assigns your sending domain and IP, based on how recipients have responded to your mail, and it decides whether future mail reaches the inbox.

Sender reputation is the running judgement each mailbox provider keeps about your sending domain and IP address. It is not one number and not published: Google, Microsoft and every other provider maintain their own, from their own users' behaviour, and none of them will tell you what it is. You infer it from outcomes.

What feeds it is recipient behaviour, weighted towards the negative. Complaints, marking as spam, and deleting without opening push it down. Bounces push it down hard, because mailing addresses that do not exist demonstrates a list you did not verify. Opens, replies, and being moved out of spam push it up, and replies count for more than opens because they are harder to fake. Volume itself is neutral; sudden changes in volume are not.

The asymmetry is what matters operationally. Reputation falls in days and recovers over weeks or months, and there is no way to appeal or reset it. This is why sending caps, verification before sending, and stopping automatically on a complaint spike are worth enforcing in code rather than leaving to judgement. By the time a person notices the problem in a dashboard, the damage is already done and the recovery is slow.

In practice:
- A complaint rate crossing 0.3%, after which Gmail begins filtering most of a domain's mail.
- A domain that sent to a stale list once and spent two months recovering its inbox placement.
- A circuit breaker that stops sending automatically when bounces exceed a threshold, rather than waiting for someone to look.

Q: How do I check my sender reputation?
A: You cannot read it directly, but you can read proxies. Google Postmaster Tools shows a domain reputation band and a spam rate for Gmail specifically, which is the closest thing to a direct reading. Beyond that, your own bounce rate, complaint rate and reply rate are the signals worth watching, plus whether inbox placement differs between providers. A drop that shows up at one provider and not others points at that provider's judgement rather than at your list.

Q: How long does it take to repair a damaged reputation?
A: Weeks at best, months if the damage was severe, and only if the sending behaviour that caused it stops. Providers weight recent behaviour most heavily, so a period of low-volume, high-engagement sending is what rebuilds standing. There is no reset and no appeal. Moving to a new domain starts a fresh reputation but also starts from zero history, which means warming up again.

Q: Does reputation follow the domain or the IP?
A: Both, tracked separately. On shared sending infrastructure the IP reputation is partly other senders' doing, which is an argument for a dedicated IP at volume and against one at low volume, where a dedicated IP simply has too little traffic to build a history. Domain reputation is yours alone and is the one that follows you between providers.

## Email verification

https://kamugrows.com/glossary/email-verification

Email verification is checking that an address exists and can receive mail before you send to it, so a message to a dead address never becomes a bounce against your domain.

Email verification is checking whether an address can actually receive mail before you send anything to it. It matters because bounces are one of the fastest ways to damage a sending domain, and every bounce is avoidable: the address either exists or it does not, and that is knowable in advance.

Verification happens in layers, cheapest first. Syntax checking catches malformed addresses. Domain checks confirm the domain resolves and publishes MX records, which rules out typos and dead companies. An SMTP check opens a connection to the receiving server and asks whether the mailbox exists without sending anything, which is the layer that catches a real domain with no such person. Most services combine these and return a confidence rather than a yes.

Two things worth knowing. Some providers accept mail for every address and bounce it later, which makes an SMTP check inconclusive; these come back as "accept-all" and are a judgement call rather than a verified address. And verification decays: people leave jobs, so an address verified six months ago is not verified now. Lists have a shelf life measured in months.

Also called: Email validation.

In practice:
- A domain with no MX record, ruled out before sending.
- An address at a real company where the person left, caught by an SMTP check.
- An accept-all domain that returns inconclusive, where sending is a risk you take knowingly.

Q: Is verification worth the cost?
A: Almost always, because the alternative is priced in reputation rather than money. Verification costs a fraction of a cent per address; a bounce rate above a few percent costs inbox placement for every future send from that domain, and recovery takes weeks. The arithmetic favours verifying even quite large lists.

Q: Can verification tell me if someone still works somewhere?
A: Only indirectly. An SMTP check tells you whether the mailbox exists, and many companies keep a departed employee's address alive as a forward, so it can pass for someone who left months ago. Job-change signals come from elsewhere, usually from professional profiles. Verification protects your domain; it does not tell you the person is still the right person.

## Ideal customer profile

https://kamugrows.com/glossary/ideal-customer-profile

An ideal customer profile is a description of the kind of organisation that gets the most value from what you sell, written specifically enough that you can tell whether a given company matches it.

An ideal customer profile describes the kind of organisation that gets the most out of what you sell. It describes a company rather than a person, which is what separates it from a buyer persona: the ICP says which companies to approach, and the persona says who inside one to write to.

The test of an ICP is whether it can exclude anything. "Small and medium businesses that want to grow" is not a profile, because no company fails it. A usable one names attributes you can actually observe from outside: size band, what they sell and to whom, the tools they run, how they got funded, and whatever situational fact makes the problem you solve urgent for them right now. That last part is what most profiles are missing, and it is the part that determines whether outreach lands.

It is built from customers you already have, not from the market you would like. The useful exercise is to look at who renews, who expands, and who reaches value fastest, then ask what those accounts share that your churned ones do not. Before you have customers, an ICP is a hypothesis, and it should be written down as one so it can be corrected rather than defended.

Also called: ICP.

In practice:
- Series A B2B SaaS companies, 20 to 100 staff, selling to engineering teams, already running a CRM but no dedicated marketer.
- Not an ICP: "fast-growing tech companies", which excludes nothing observable.
- A profile narrowed after noticing that every account which churned had bought without a technical owner.

Q: How is an ICP different from a buyer persona?
A: An ICP describes the organisation and a persona describes a person inside it. The ICP decides which companies are worth approaching at all; the persona decides who within a matching company to write to and what they care about. You need both, and they answer different questions: getting the ICP wrong means writing to the wrong companies, and getting the persona wrong means writing to the wrong person at the right company.

Q: How narrow should an ICP be?
A: Narrow enough that a majority of companies fail it, and no narrower than your evidence supports. Early on the temptation is to keep it wide so as not to rule out revenue, which produces outreach that is relevant to nobody. The practical test is whether the profile changes what you write: if the same email works for everything inside it, the profile is still too broad.

Q: Can I build an ICP before I have customers?
A: You can write a hypothesis, and you should, but call it that. Pre-revenue the profile is a guess about who has the problem badly enough to pay, and its value is that it is specific enough to be proven wrong quickly. Treat the first fifty conversations as the experiment that corrects it rather than as validation of it.

## Lead enrichment

https://kamugrows.com/glossary/lead-enrichment

Lead enrichment is adding context to a bare name, email or domain, such as company size, role, funding or the tools in use, so you can judge whether the lead fits your profile.

Lead enrichment is turning a thin record into one you can act on. You start with a domain, or a name and an email, and you add the attributes that let you judge fit: headcount, what the company sells, the person's role and seniority, funding history, the technologies on their site, recent hiring.

It exists because qualification is otherwise manual. A list of two thousand domains cannot be assessed by a person in reasonable time, but it can be filtered mechanically once each row carries the attributes your profile is written in terms of. Enrichment is what makes an ICP operational rather than aspirational.

The important limitation is that enrichment data is inferred and dated. Headcount comes from professional profiles that people do not update, technology detection reads what is on the public site and misses anything internal, and job titles mean different things at different companies. Treat a field as evidence rather than fact, weight recency, and be careful about the ones that drive automated decisions. A message that opens by naming a role the person no longer holds is worse than one that names nothing.

Enrichment is also processing of personal data. Under GDPR the person has a right to know where their data came from, which means keeping the provenance rather than flattening every source into one record.

Also called: Data enrichment.

In practice:
- A domain enriched to: 45 staff, Series A, sells to marketing teams, running HubSpot and Segment.
- A stale record showing a person as a manager who has since become a director elsewhere.
- Technology detection that misses an internal tool because it never appears on the public site.

Q: How accurate is enrichment data?
A: Good on stable firmographics, weaker on anything that changes. Domain, industry and rough size band are usually right. Job titles and headcount drift constantly because they depend on people updating profiles they have no reason to update, and roughly a third of business contacts change role each year. Technology detection sees only what is publicly visible. The practical response is to weight recency and to avoid making a claim in an email that depends on a field being current.

Q: Is enriching data on people legal under GDPR?
A: It can be, and it is regulated. Enrichment is processing personal data, so it needs a lawful basis, normally legitimate interest for business contacts, and it comes with obligations: the person can ask what you hold, where it came from, and can object. Because you obtained the data indirectly rather than from them, you also have to tell them the source. Keeping provenance per field rather than a merged record is what makes that answerable. This is not legal advice.

## Buying signal

https://kamugrows.com/glossary/buying-signal

A buying signal is an observable event that suggests a company has the problem you solve right now, which makes it the reason to write to them today rather than at some point.

A buying signal is an observable event suggesting a company needs what you sell now rather than in principle. Fit tells you who could buy; a signal tells you when, and timing is most of why outreach works or does not. The same email to the same person is welcome the week they start looking and ignored the month before.

Signals differ in how much they are worth. The strongest are specific and public: a job posting describing the exact problem you solve, a funding round that releases budget, a migration announcement, a new executive who owns the area and needs an early win, a public complaint about a competitor. Weaker ones are generic company growth or a page visit, which say something is happening but not what. The strength of a signal is roughly how well it explains why this week.

The operational value is that a signal gives you something true to open with. An email that references a specific hire or announcement is doing work no template can, because it demonstrates the message was written after looking rather than assembled from a list. That is also the failure mode: a signal cited wrongly or six months late is worse than no signal, because it shows the looking was automated and careless.

Also called: Trigger event, Intent signal.

In practice:
- A job posting for a role whose responsibilities are the problem your product removes.
- A funding round announced last week at a company that already matches your profile.
- A new VP of engineering, three weeks in, who owns the system you replace.
- A weak signal: the company grew headcount this quarter.

Q: What is the difference between a buying signal and intent data?
A: A buying signal is a specific observable event, such as a particular job posting or funding round. Intent data usually means aggregated behavioural inference sold by a vendor, such as a claim that accounts at a company have been researching a category. Signals are verifiable and explain themselves; intent data is probabilistic and often cannot be traced to anything you could reference in an email without sounding like surveillance.

Q: How quickly does a signal go stale?
A: Days to a few weeks for most, and the decay is steep. A funding round is interesting for a month and old news after that. A job posting is live until it is filled. Referencing a stale signal actively hurts, because it tells the recipient the message was generated from a list rather than written after paying attention. If you cannot act within the window, it is better to write without the signal than to cite it late.

## Reply rate

https://kamugrows.com/glossary/reply-rate

Reply rate is the share of delivered emails that get a human response. It is the most reliable measure of cold outreach, because unlike opens it cannot be faked or miscounted.

Reply rate is replies divided by emails delivered. It is the number worth managing in cold outreach, for two reasons: it cannot be inflated the way open rate can, and it is the input mailbox providers weight most heavily when deciding where to put your next message.

Open rate stopped being trustworthy when privacy proxies began pre-fetching tracking pixels, which registers an open nobody performed. Depending on the audience a meaningful share of reported opens are machine artefacts, so a campaign can show healthy opens and no human interest. A reply requires a person to have read the message and decided to respond, which is why it is both harder to move and worth more.

What counts as good depends on how narrowly you are targeting. Broad sending to a bought list runs well under 1%. Carefully aimed outreach to a tight profile, referencing something specific about each recipient, commonly runs 5% to 15%, and a very narrow campaign into a strong signal can go higher. If a campaign is under about 2%, the problem is nearly always relevance or timing rather than wording, and sending more of it makes things worse: low reply rates and rising complaints are the same underlying fact seen from two sides.

Count positive and negative replies separately. A clear no is a useful outcome and a healthy sign the message reached the right person; counting it alongside interest hides whether anything is working.

In practice:
- 12% on a hundred emails to companies that had just posted a matching job.
- 0.4% on four thousand emails to a bought list, with complaints rising through the send.
- A 6% reply rate where half the replies are declines, which is still a working campaign.

Q: What is a good cold email reply rate?
A: For carefully targeted business-to-business outreach, 5% to 15% is a normal band, and above that usually means either a very narrow list or a strong timing signal. Under 2% indicates the list or the timing is wrong rather than the copy. Broad sending to unverified lists typically comes in under 1%, which is why volume strategies need enormous numbers to work at all and damage the sending domain while doing it.

Q: Why not measure open rate?
A: Because it is no longer a measurement. Privacy features that pre-fetch images register opens that no person performed, so the number includes machine activity you cannot separate out. It can still be read as a rough directional signal on subject lines within a single audience, but it should not be used to judge whether a campaign is working, and it should never be used to decide whether to scale one.

Q: Should I count a negative reply?
A: Count it, and keep it separate. A no means the message reached a real person who read it, which is information about deliverability and targeting both. Tracking positive and negative separately tells you whether a low reply rate is a reach problem or an interest problem, and those have completely different fixes.

## AI agent

https://kamugrows.com/glossary/ai-agent

An AI agent is a system where a language model pursues a goal by taking actions with tools, deciding its own next step, rather than producing text in response to each prompt.

An AI agent is a system in which a language model works towards a goal by taking actions, rather than by answering one prompt at a time. The model is given tools it can call, some memory of what it has already done, and a loop: choose an action, observe the result, decide what to do next, stop when the goal is met.

What separates an agent from a chatbot is who decides the next step. In a chatbot a person supplies each turn and the model responds. In an agent the model chooses the sequence itself, which means it can do work while nobody is watching, and it means the choices it makes are consequential in a way a wrong answer in a chat is not. An agent that can send email can send the wrong email.

That consequence is why the interesting engineering in an agent is not the model. It is the boundary: which tools exist at all, what limits are enforced in code rather than requested in a prompt, which actions require a person to approve them, and whether every action is recorded with the reason behind it. A prompt asking a model to be careful is not a control. A hard cap on how many messages can be sent in a day is.

In practice:
- An agent that reads a codebase, edits files, runs tests, and iterates until they pass.
- An agent that researches prospects, writes to them, and answers replies within a daily cap it cannot exceed.
- Not an agent: a model that drafts an email and stops, waiting for a person to send it.

Q: What is the difference between an AI agent and a chatbot?
A: Who decides the next step. A chatbot responds to prompts a person supplies, one turn at a time. An agent picks its own next action in a loop, uses tools to affect things outside the conversation, and can run with nobody present. The practical consequence is that an agent's mistakes are actions rather than sentences, which is why agents need enforced limits and audit trails that a chatbot does not.

Q: What stops an agent doing something harmful?
A: Only the boundary you build. Instructions in a prompt are a preference, not a control, because the model can misread them and there is nothing to stop it if it does. Real controls are structural: not giving it a tool for the thing it must never do, enforcing caps in code so exceeding them is impossible rather than discouraged, requiring human approval for anything that spends money or is hard to reverse, and logging every action so a wrong one can be found and undone.

Q: Do agents need to be supervised?
A: They need supervision at the right points rather than continuously, which is the whole design problem. Watching every step removes the reason to use one; watching nothing means discovering mistakes after they have compounded. The workable arrangement is that reversible, bounded work proceeds unattended and anything expensive, public or irreversible waits for a person.

## Growth agent

https://kamugrows.com/glossary/growth-agent

A growth agent is software that performs marketing and sales work on its own schedule, deciding what to do next within hard limits rather than presenting you with tools to do that work yourself.

A growth agent is software that does the marketing and sales work rather than helping you do it. The distinction is not the presence of AI, which most marketing tools now claim. It is whether the work happens when nobody opens the product.

A marketing tool waits. It gives you a campaign builder, a sequence editor, a content calendar, and it produces nothing until a person sits down and drives it. That is the right shape when there is someone whose job is to drive it, and it is the wrong shape for a founder who intended to spend the day building. The tool is not the bottleneck; the attention is.

A growth agent inverts that. It reads your product and market, decides what to work on, and does it on a schedule of its own: finding people who fit, writing to them, answering replies, drafting content, posting. The founder's job becomes approving the calls that need judgement and reading what happened, rather than operating the software.

What makes that trustworthy is the same boundary any agent needs. Limits are numbers you set and the agent cannot exceed. Anything that spends money, publishes publicly, or is hard to undo waits for approval. Every action is recorded with the reason behind it, so the work can be checked after the fact rather than watched during it. Without those, autonomy is just an unattended process with your domain reputation at stake.

In practice:
- Software that finds and writes to twenty matching prospects a day without being opened.
- An agent that drafts a video ad, builds the campaign, and leaves it paused for a person to start.
- Not a growth agent: a sequence tool with an AI writing assistant, which still needs someone to run it.

Q: How is a growth agent different from marketing automation?
A: Marketing automation executes rules you write: if this happens, send that. The thinking is yours and encoded in advance, and the tool does the repetitive part. A growth agent decides what to do, including things you did not specify, within the limits of the plan and the objectives you activate. The trade is control for attention: automation does exactly what you told it and nothing more, and an agent does work you never got round to specifying, which is more useful and requires more care about what it is allowed to do.

Q: What happens if it writes something wrong?
A: That depends entirely on what was made reversible. It is why the useful arrangement puts anything public, expensive or hard to undo behind human approval, and lets only bounded, reversible work proceed alone. A draft nobody sent costs nothing. An email to two hundred people cannot be recalled, which is why sending caps and suppression lists belong in code rather than in a prompt, and why every action needs a log entry explaining why it was taken.

Q: Does a growth agent replace hiring a marketer?
A: It changes what you would hire for rather than removing the need for judgement. The everyday production work of finding people, writing to them and drafting content is what an agent does without getting tired. Positioning, deciding what the company is for, and the calls that depend on knowing things not written down anywhere are not that. The honest comparison is with the first growth hire a small team cannot yet justify, not with an experienced head of marketing.

## Human in the loop

https://kamugrows.com/glossary/human-in-the-loop

Human in the loop means an automated system pauses at defined points for a person to approve, edit or reject what it proposes, rather than either running unattended or being driven manually throughout.

Human in the loop describes an arrangement where an automated system does the work but stops at defined points for a person to decide. It sits between full automation, where nobody sees anything until afterwards, and manual operation, where a person drives every step.

The design question is not whether to have a person involved but where. A checkpoint on everything is the same as having no automation, because the bottleneck moves to the review queue and the person starts approving without reading, which is worse than no checkpoint at all: it manufactures the appearance of oversight. A checkpoint on nothing means finding out about a mistake after it has been repeated two hundred times.

The useful rule is to place checkpoints by reversibility and blast radius rather than by importance. Work that is cheap, private and undoable should proceed unattended: a draft, a piece of research, a list of candidates. Work that spends money, publishes to an audience, contacts someone for the first time at scale, or cannot be recalled should wait for a person. The second category is small, which is what keeps the review load survivable and the reviews genuine.

A checkpoint is only real if declining is as easy as approving and the system does not proceed on silence. A prompt that defaults to yes after a delay is automation wearing a review as a costume.

Also called: HITL.

In practice:
- An agent that drafts and sends routine replies alone but escalates anything mentioning pricing.
- A video ad campaign built and left paused, so starting it is a deliberate human act.
- A bad checkpoint: forty drafts a day in a queue, approved unread by Friday.

Q: Where should the human checkpoint go?
A: On actions that are irreversible, cost money, or reach an audience. Reversibility is the better criterion than importance, because an important decision that can be undone tomorrow is safer to automate than a trivial one that cannot. In practice that means drafting, research and internal work run unattended, while publishing, sending at volume, and spending wait for a person.

Q: Does human in the loop slow everything down?
A: Only if the checkpoints are in the wrong places. Reviewing everything creates a queue that becomes the constraint and, worse, trains the reviewer to approve without reading. Reviewing the small set of consequential actions costs a few minutes and is the part that makes the rest safe to run alone. If the review load feels heavy, the answer is usually fewer checkpoints on more clearly bounded work, not faster reviewing.

## Generative engine optimization

https://kamugrows.com/glossary/generative-engine-optimization

Generative engine optimization is the practice of making a page likely to be quoted and cited by AI answer engines, which summarise sources rather than sending a reader a list of links.

Generative engine optimization is making your content likely to be used and cited when an AI system answers a question. It matters because the shape of the result changed: an answer engine reads several sources and writes one response, naming a few of them. There is no page two, and being the fourth-best source often means not appearing.

Most of it is ordinary quality work, done with a different reader in mind. A model extracting an answer wants a passage that is true standing completely alone, because it will be lifted away from the page around it. That favours naming the subject rather than pronouning it, answering the question in the first sentence rather than building to it, and stating specifics a summary cannot blur. Definitional and comparative pages do disproportionately well, because they answer a question someone actually asked in a form that can be quoted.

The technical part is smaller than vendors suggest but not nothing: let AI crawlers reach the page, keep the substance in the HTML rather than behind interaction, so an accordion that holds one answer in the DOM at a time is a real problem, mark up structure so a machine can tell a definition from a navigation label, and describe your organisation consistently enough that a model resolves it to one entity.

What does not work is the thing SEO taught people to try. There is no keyword density to tune and no volume play, because there is no ranked list to climb. Thin pages produced at scale are worse than useless here: a model choosing between sources is choosing which to trust, and a page that says less than the others is simply not chosen.

Also called: GEO, AI search optimization.

In practice:
- A glossary entry answering the question in its first sentence, quoted verbatim by an answer engine.
- An llms.txt file that gives models the definitions themselves rather than only a list of links.
- A page that loses citations because its answers sit inside an accordion that renders one at a time.

Q: How is GEO different from SEO?
A: SEO competes for position in a list of links; GEO competes to be the source an answer is written from. That changes what wins. Ranking rewards matching a query and accumulating authority, so there is room for the fourth result. Citation rewards being the clearest, most self-contained, most trustworthy passage on the point, and there is usually no room for fourth place. The foundations overlap almost entirely, since a page a crawler cannot read cannot be cited either.

Q: Does an llms.txt file actually help?
A: It is cheap and plausibly useful, and it is not a ranking mechanism. The convention is a plain-text file describing your site for language models. No major provider has committed to reading it, so treating it as a lever would be overclaiming. What makes it worth writing is that generating it from the same source as your pages forces the site's claims into one consistent form, and a version carrying the actual definitions rather than only links gives a model substance it can use without parsing HTML.

Q: Can I make an AI recommend my product?
A: Not directly, and anyone selling that is selling something else. A model recommends what its sources support, so the work is making the sources say something true and specific about you: clear pages about what the product does, consistent naming so the entity resolves, and corroboration off your own site, since a model weights independent mentions more heavily than your marketing. What you influence is whether you are findable and describable. What gets recommended is decided by the evidence.

## AI SDR

https://kamugrows.com/glossary/ai-sdr

An AI SDR does the sales development job without a person doing it: finding people who match your buyers, writing the first message, and handling what comes back.

A sales development representative is the person at the top of a sales team whose job is to start conversations. They build a list of people worth contacting, research enough about each to write something that is not obviously a template, send it, and hand over anyone who replies with interest. It is the least glamorous job in sales and the one most companies fill first, because nothing further down the funnel happens without it.

An AI SDR is software that does that job rather than helping somebody do it. The distinction matters and is frequently blurred in marketing: a tool that drafts emails faster is not an AI SDR, it is a faster tool for an SDR. The label properly belongs to something that decides who to contact, decides what to say, sends it, and deals with the reply, without waiting for a person at each step.

The category is genuinely new and the claims are uneven. Most products described this way still need somebody to define the audience, approve the copy, or press send, which is the part that consumes the time the product is supposed to give back. The useful test is what happens if nobody opens the tool for a week: a real one has been working, and most have not.

It is also a narrower label than it first appears. Sales development is the top of the funnel and nothing else. An AI SDR is not doing your content, your landing page, your ads or your listings, and a product that does those things is doing more than the term describes. Kamu is usually filed under this heading, and outreach is one of six objectives it runs.

Also called: AI sales development representative, AI SDR agent, autonomous SDR.

In practice:
- Finding twenty people this week whose companies just shipped something your product would have helped with
- Opening an email with a specific thing that person wrote, rather than a compliment about their company
- Reading a reply that says "not right now" and closing the thread instead of sending a follow-up
- Stopping permanently when somebody asks to be taken off the list, across every product the same company runs

Q: Is an AI SDR the same as a growth agent?
A: No, and the difference is scope rather than quality. An AI SDR does sales development: prospects, first messages, replies. A growth agent covers more of the job, including content, social, a landing page, listings and sometimes paid ads. Outreach is one part of it. Most products called AI SDRs really are only doing the top of the funnel, which is why the narrower label sticks.

Q: Does an AI SDR replace a human SDR?
A: It replaces the volume half. What it does not replace is the judgment: reading a room on a call, deciding a lead is worth an exception, noticing that everybody is objecting to the same thing and telling the founder the pricing is wrong. Teams that keep both tend to put the software on the first touch and the person on anything that has already replied.

Q: How do you tell a real one from a tool with an AI label?
A: Ask what happens if nobody opens it for a week. A real one has been finding people, writing to them and answering replies the whole time. A tool with an AI label has been waiting, because somewhere in its flow there is a step that needs a person to approve, define or press something.

Q: Is cold email from an AI SDR legal?
A: It is subject to exactly the same rules as any other cold email, and the software makes no difference to them. In practice that means a real sender address, a working unsubscribe honoured immediately and permanently, a postal address in the footer, and under GDPR a legitimate-interest basis you could actually defend. Anything that treats those as optional is a deliverability problem before it is a legal one.

## Cost per qualified lead

https://kamugrows.com/glossary/cost-per-qualified-lead

Cost per qualified lead is everything you spent on finding customers divided by the number of people who turned out to be worth talking to. It is the number that tells you whether outreach paid for itself.

Add everything you spent trying to find customers over a period: software, data, sending costs and the hours if you are being honest about your own time. Divide that by the number of people who came out of it worth a real conversation. That is cost per qualified lead, and it is the figure that decides whether a channel keeps running.

The difficulty is entirely in the word qualified, and it is why published figures are almost never comparable between two companies. One counts anybody who replied. Another counts anybody who replied without saying no. A third counts only people who took a call, and a fourth only those who reached a proposal. Those four definitions can differ by a factor of ten on identical activity, so a cost per qualified lead quoted without its definition is a number wearing a suit.

The second difficulty is what goes in the numerator. Tool subscriptions are easy and everybody includes them. Data and enrichment credits are usually included. The founder's own hours almost never are, which is the single largest reason the figure looks better for do-it-yourself outreach than it is: four hours a week researching prospects is not free, it is the most expensive input in the whole calculation and the one nobody invoices for.

The number is also useless early. Twenty emails and one reply is a cost per lead of whatever you spent, and it tells you nothing, because one reply is one person rather than a rate. It becomes meaningful somewhere in the hundreds of sends, which is the same reason serious benchmarks carry a sample size and the unserious ones do not.

Also called: CPQL, cost per qualified lead, cost per SQL.

In practice:
- €49 of software and 300 emails producing 4 people worth talking to, for about €12 each before counting your own time
- The same 300 emails costing four hours a week of founder research, which at any sensible valuation dwarfs the subscription
- A figure quoted by a vendor that counts every reply as qualified, including the ones that said no

Q: What counts as qualified?
A: Whatever you decide, write it down and keep it. The most defensible line for a small company is somebody who replied, is a plausible fit, and has not said no. You can check that from the thread instead of relying on an opinion about intent. What matters far more than where you draw the line is that you stop moving it, since a definition that drifts makes every month incomparable with the last.

Q: Should I include my own time?
A: Yes, and it is usually the largest input. Outreach done by hand looks cheap because the hours are not invoiced, and that is exactly what makes the comparison between doing it yourself and paying somebody misleading. Put an honest hourly figure on your own time and the arithmetic often reverses.

Q: What is a good number?
A: There is no cross-industry answer worth repeating, because it moves with contract value more than with anything about the outreach. A hundred euros a lead is excellent if a customer is worth twenty thousand and ruinous if they are worth two hundred. The only comparison that means anything is against your own previous months, on an unchanged definition.

Q: Why do published benchmarks vary so much?
A: Because qualified is undefined, the numerator rarely includes labour, and most figures are published by somebody with an interest in how they look. When you see one, ask the sample size and the definition first: a rate without either is not evidence, it is decoration.

## AI marketing agent

https://kamugrows.com/glossary/ai-marketing-agent

An AI marketing agent is software that decides what marketing work to do next and does it on its own schedule, inside limits a person set, instead of waiting for a person to operate it.

An AI marketing agent is software that does marketing work on its own schedule. It reads the product and the market, decides what is worth doing today, does it, and writes down why. The person who owns it sets the limits and reads the log. The distinction is about who is driving, and it has little to do with how much AI is inside.

Asked "what is a marketing agent" with no AI in front, a job board answers with a role: a person or a firm hired to market on someone else's behalf. This page is about the software sense.

## How it differs from automation and from a writing assistant

Marketing automation runs rules a person wrote in advance. When a trial starts, send this sequence; when a lead opens twice, add a tag. The thinking happened once, at setup, and the tool repeats it. An AI writing assistant sits at the other end: it thinks on demand and never acts. Somebody asks for a post, gets a draft, edits it and presses publish.

An agent takes neither seat. It chooses the next action from a plan and the current state of things, carries it out unasked, and comes back to a person only for what it is not allowed to decide. The useful test is a week with nobody logged in: automation has done exactly what it was told, an assistant has done nothing, and an agent has been working.

## What one does all day, and where a person still decides

It finds people who match the buyer profile and checks that each address exists before writing, then drafts a first email and, later, a reply. Another day it picks a search phrase with buyers behind it and writes the article, or turns what shipped this week into a post for each channel. A landing page gets rewritten once a version has been measured long enough to judge. Each is a small decision followed by a small action, and the value is in the accumulation: none of it would have happened while the founder was building.

Two things stay with a person. Money: a budget is set by a person, and an agent that could raise its own would be a liability. And anything that cannot be undone: a campaign going live or a message to two hundred people belongs behind an approval, since there is no recall button. Positioning sits with the person too, because what the company is for depends on things no crawler can read. The workable arrangement gives the agent the bounded, reversible work and routes the rest to a decision inbox.

## How to judge one

Three questions sort the field. Does it act when nobody is watching, or does every flow end at a button someone has to press? Does it log the reason for each action, so the work can be checked afterwards rather than supervised during? And what does it refuse: a real agent carries caps and suppression lists in code, and will say what it will not do.

### Where Kamu fits

Kamu is an AI marketing agent for SaaS founders. Given a product URL it reads the site and the market, writes a plan, then works every day on cold email, search articles, social posts, a landing page, directory listings and video ads. Anything that spends money or changes the plan waits for the founder's approval in chat, Slack, Telegram or WhatsApp. Every action is logged with its reason, unsubscribes are honoured across every product in the account, and ad campaigns are created paused.

Also called: marketing agent, AI marketing employee, autonomous marketing agent.

In practice:
- Software that, on a Tuesday when nobody logged in, found twelve people matching the buyer profile, verified their addresses and wrote to six of them within its daily cap.
- An agent that reads a product's GitHub releases, turns the week's shipping into a LinkedIn post and an X post, and says nothing on a week where nothing shipped.
- Not an AI marketing agent: a campaign tool with an AI button that drafts the email, which still needs someone to build the list and press send.

Q: Is an AI marketing agent the same as marketing automation?
A: No. Marketing automation runs rules a person wrote in advance and does nothing that was not specified. An AI marketing agent decides what to do from a plan and the current situation, including work nobody wrote a rule for, and it does that within limits set in advance. Automation is exact and cannot surprise you. An agent can, which is why the limits it works inside matter more than its cleverness.

Q: What is a marketing agent in the job sense?
A: A person or an agency hired to do marketing for a company that does not employ its own marketers, paid a fee or a commission. Job boards use the phrase that way, and the listings under it are for people. The software sense is newer: a program that does the marketing work itself, on a schedule, and reports back to whoever owns it.

Q: What should an AI marketing agent refuse to do?
A: Anything that spends money or cannot be undone without a person approving it, and a set of things it should never do at all: write to someone who unsubscribed, send to an unverified address, exceed a daily cap, start an ad campaign live, invent a testimonial or a rating. The refusals belong in code rather than in a prompt, because a prompt can be argued with and a function cannot.

Q: Does an AI marketing agent replace a marketer?
A: It replaces the production half of the job: finding people, writing first drafts, posting what shipped, filing listings. Positioning, pricing and the calls that depend on knowing things not written down anywhere stay with a person. For a small SaaS the fair comparison is with the first marketing hire the company cannot yet afford, since that is the work an agent takes on.

## SaaS marketing

https://kamugrows.com/glossary/saas-marketing

SaaS marketing is the work of getting a subscription product found, tried and kept, where the first payment starts the relationship instead of ending it.

SaaS marketing is the work of getting a subscription product found, tried and kept. It differs from marketing a one-off purchase in one way that changes everything downstream: the sale is the start. A customer who signs up in March and leaves in May has paid for two months and cost more than that to acquire, so the job is unfinished when the card goes in.

## What is different about marketing a subscription

Churn is the score. A channel that brings in people who leave within six weeks looks fine on the signup chart and is quietly losing money, so retention is read next to acquisition from the first month.

A trial does the selling. Almost every SaaS is tried before it is bought, so the marketing aims at getting someone to a first useful moment inside the product rather than at closing them on a call. Onboarding emails and the first screen a new account sees are marketing surfaces, though nobody in marketing usually owns them.

The pricing page is content. It is the most read page after the homepage, it is where a buyer decides whether this is meant for a company like theirs, and it ranks for the product's name plus "pricing".

Early on, the founder is the marketer. Nobody else yet knows the product or the buyer well enough to write about either, and a founder's name gets replies that a company account does not.

## A SaaS marketing strategy for a small team

Some channels work before there is a brand, and some do not. Search works early because the phrases a buyer types ("X alternative", "how to do Y") are specific and the pages that rank for them are often weak. Cold email works early because it needs no audience, only a clear picture of who buys and a real reason to write to them. Communities work if the founder answers questions there for months before mentioning the product. Directories bring a backlink and a page where buyers compare. A public changelog is the cheapest content there is: the work was already done.

Paid acquisition at scale and brand campaigns tend not to work yet. Paid needs a conversion rate that has been measured rather than guessed and a landing page that has been through a few versions. Brand needs years of repetition and a budget a small team does not have.

Sequence follows. Start with cold email and one landing page, because they produce replies inside a week and the replies teach the positioning. Add search content once the phrases are known, since articles take months to rank and should target phrases real buyers used in those replies. List the product in directories the same month. Try a small paid test only once the trial converts at a rate you can state.

## How to measure it

Replies first, then trials, then accounts still paying at ninety days. Impressions and followers are what a channel reports about itself. A reply to a cold email is a buyer saying the problem is real; a trial started says the pitch was clear enough to try. Only the account still paying at ninety days is revenue. Cost per qualified lead ties them together: spend divided by the number of people worth talking to.

Kamu runs the early channels in this list for a founder: cold email one person at a time, search articles for phrases with buyers behind them, directory listings, and a landing page rewritten when the numbers say to. Paid ads are off by default, being the one job that spends money.

Also called: software as a service marketing, B2B SaaS marketing.

In practice:
- A two-person team that spends its first quarter on cold email and one landing page, then writes search articles against the phrases buyers used in their replies.
- A pricing page rewritten with a plan named for the buyer ("for agencies") after three replies in a month asked whether the product was meant for them.
- Not SaaS marketing in the sense used here: a one-off software licence sold with a launch campaign and no follow-up, where churn is a number nobody tracks.

Q: What is the difference between SaaS marketing and B2B marketing?
A: B2B marketing is defined by who buys: a company rather than a consumer. SaaS marketing is defined by how the product is sold: as a subscription, usually with a trial, where the customer can leave any month. Much B2B software is both. The subscription is what changes the job, since it makes retention part of marketing and turns the trial into the main sales conversation.

Q: Which marketing channels work first for a new SaaS?
A: Cold email and one landing page, because they produce replies within a week and cost time rather than money. Search content next, written against the phrases buyers actually used in those replies, since articles take months to rank. Directory listings alongside, for the backlinks. Paid acquisition last, and only once the landing page converts at a rate that has been measured.

Q: How do you measure SaaS marketing?
A: By replies, trials started and accounts still paying after ninety days, rather than by impressions or followers. The first two say whether the message lands with a stranger. The third says whether the channel brought people who stay, which is the only one of the three that is revenue. Cost per qualified lead ties spend to the number of people who were worth talking to.

Q: Does a SaaS need a marketing hire before it has customers?
A: Usually not. Before there are customers the work is finding out who buys and why, and the founder is the only person who can change the product and the price in the same conversation. A first marketing hire tends to work better once the positioning has stopped moving and there is a repeatable channel for them to run.

## Founder-led sales

https://kamugrows.com/glossary/founder-led-sales

Founder-led sales is the stage where the founder finds the buyers, runs the conversations and closes the deals in person, before the company has anyone whose job that is.

Founder-led sales is the founder doing the selling: finding the buyers, writing the first message, taking the calls, agreeing the price and chasing the invoice, with no salesperson in between. Nearly every software company starts this way, and the ones that skip it tend to come back to it after the first sales hire fails.

## Why it works early

The founder can change the product and the price in the same call. A buyer says the integration they need is missing, and the person they are talking to can promise it for next month and mean it. A buyer balks at the price, and the person on the call can invent a plan on the spot. No salesperson has that authority, and no sales deck survives contact with it.

The founder also hears objections raw. A salesperson reports "pricing concerns"; the founder hears the exact sentence, notices that four people this month said the same thing, and fixes the positioning on Friday. That loop between what buyers say and what the company does is the fastest it will ever be, and the point of the stage is to run it as many times as possible.

Buyers reply to founders. A first email from the person who built the thing gets answered where the same email from a sales development rep does not, because the recipient can tell the difference and knows which of the two can actually help.

## When it stops working

It stops when the founder runs out of hours. A founder who spends the morning building and the afternoon on calls is doing two jobs at half speed, and once there are more replies than one person can answer carefully, the finding and the first message start to slip. The pipeline dries up three weeks later, and nobody notices why until it has.

It also stops when the sale is understood. The stage exists to learn who buys, why, and what they say when they hesitate. Once those answers repeat, the founder is no longer learning on calls, only closing on them, and the time is worth more elsewhere.

## What to hand off first

Hand off the finding first, then the first email, and keep the closing call longest. Prospecting is the part that eats time and teaches least: a list of matching companies with verified addresses is the same list whoever built it. The first message is close behind, once the founder has written enough of them to know what works. The reply, the call and the price stay with the founder, because that is where the product and the positioning still get changed.

An agent that finds and writes to buyers under the founder's name sits in that first slot. Kamu works out who buys the product, finds people who match with a real address, writes the first email in the founder's voice and drafts a reply when someone answers. The founder reads the reply and takes the call. The judgment stays with the person, and the finding and the first email come off the calendar.

Also called: founder-led selling, founder sales, founder led sales.

In practice:
- A technical founder who takes every demo call for the first year, then rewrites the pricing page after noticing that the same objection came up in nine of them.
- A founder's cold email that opens with the buyer's own changelog entry and gets a reply the same afternoon, where the identical message from a sales rep would have been ignored.
- Not founder-led sales: a founder who hires two sales reps before the first ten customers and hands them a deck, so nobody in the company has yet heard a buyer hesitate.

Q: How long should founder-led sales last?
A: Until the answers stop changing. When the founder can predict who will buy, what they will object to and what they will pay, the calls have stopped teaching anything, and that is the signal to hand parts of the job to someone or something else. In practice that lands somewhere between the first ten and the first hundred customers, and it depends more on how varied the buyers are than on the calendar.

Q: What is the first thing to hand off in founder-led sales?
A: The prospecting: finding the people who match and getting a verified address for each. It takes the most hours and depends least on the founder's judgment, since a list of matching companies is the same list whoever builds it. The first email goes next, once the founder has written enough of them to know what gets a reply. The conversation itself stays with the founder longest.

Q: Can software do founder-led sales?
A: It can take the part that is not judgment. An agent can work out who buys, find people who match, verify their addresses and write a first email in the founder's voice under the founder's name, and it can draft the reply. It cannot change the product or the price on a call, which is the thing that makes the stage work, so the useful arrangement gives it the finding and the first message and leaves the founder the conversation.

Q: Is founder-led sales the same as founder-led growth?
A: Founder-led sales is the narrower term: the founder personally selling, one conversation at a time. Founder-led growth is sometimes used for everything a founder does to acquire customers, including writing, speaking and building an audience in public. The sales version is the one with a natural end, because at some point there are more buyers than one person's calendar can hold.

## Outbound sales

https://kamugrows.com/glossary/outbound-sales

Outbound sales is starting conversations with buyers who have not come to you, by working out who is likely to need the product and contacting them directly by email, LinkedIn or phone.

Outbound sales is going to the buyer instead of waiting for the buyer to arrive. A company works out who is likely to need what it sells, finds those people, and contacts them directly. The word describes the direction of the first message: it travels from the seller outwards, to somebody who did not ask for it, which is what separates it from a lead who filled in a form.

## The channels and the roles

Email is the main channel because it scales and can be aimed. A cold email to a named person, referencing something specific about their company, is still the cheapest way to open a business conversation. LinkedIn adds a second route to the people who never open email, through a connection request and a short message. Phone calls are the oldest channel and the most expensive per conversation, and still work in markets where the buyer picks up.

In a sales team the work is split in two. A sales development representative, the SDR, builds the list and sends the first messages, then hands anyone who replies with interest to an account executive, the AE, who runs the calls and closes. At a small SaaS the founder is both, which is the main reason outbound feels expensive there.

## Inbound vs outbound sales

Inbound is the buyer arriving on their own, through a search, a referral, a directory or a piece of content, and asking to talk. Outbound is the seller arriving first. Inbound leads convert better because the buyer has already decided the problem is worth solving, and once the content exists they cost less per conversation. Outbound gives control over who you talk to: a company that wants logistics firms in Germany can write to exactly those, while inbound brings whoever happened to search.

The two usually run together. Inbound takes months to build and outbound produces conversations within a week, so early companies lean on outbound while the inbound engine is built, and use what buyers say in outbound replies to decide what the inbound content should be about.

## What it costs, when to start, and how to judge it

For a small team the cost is mostly time. Building a list of people who match, finding a real address for each, writing a message that is not a template and following up takes most of a day for a modest batch, every week, and the founder's day is the expensive one. Tools for finding addresses and sending add a monthly fee, and a new sending domain needs weeks of warmup before it can carry any volume.

A SaaS should start outbound as soon as it can say who buys and why, which is earlier than founders tend to think. The first ten conversations are worth more as research than as revenue, because they show whether the pitch lands with a stranger who owes you nothing.

Judge it by reply rate first, then by the number of qualified conversations, and finally by cost per qualified lead. Messages sent and open rates say nothing about whether anyone wanted to talk.

The founder-scale version is small and specific: a handful of researched emails a day, sent from the founder's own domain, with every reply read by the founder. Kamu does the first half of that for a founder, finding people who match, verifying each address before sending, writing one email at a time within a daily cap that comes with the plan, and drafting the reply when someone writes back.

Also called: outbound, outbound prospecting, outbound selling.

In practice:
- Twenty emails a week to heads of operations at logistics companies that posted a job for a dispatcher last month, each one naming the job post.
- A LinkedIn connection request and a two-line note to a founder who wrote publicly about a problem your product handles.
- Not outbound: answering a demo request that came in from a search result, however fast you answer it.

Q: What does outbound sales mean?
A: Outbound sales means the seller makes the first move: working out who is likely to need the product, finding those people, and contacting them by email, LinkedIn or phone before they have shown any interest. The opposite is inbound, where the buyer arrives through search, a referral or content and asks to talk. The word refers to the direction of the first message.

Q: What is the difference between inbound and outbound sales?
A: Who moves first. In inbound the buyer comes to you, usually through search or a referral, and has already decided the problem is worth solving, so those conversations convert well and cost little once the content exists. In outbound you go to the buyer, which converts less often and lets you choose exactly who you talk to, and it produces conversations within a week rather than months.

Q: When should a SaaS start outbound sales?
A: As soon as it can describe who buys and why, and before the founder feels ready. The first conversations are research more than revenue: they show whether a stranger understands the pitch and what they object to, and that is cheaper to learn from ten emails than from a launch. The prerequisites are a sending domain that has been warmed up and a clear reason each person is being written to.

Q: How much does outbound sales cost a small team?
A: Mostly time, and then a few tools. Researching a list, finding verified addresses, writing messages that are not templates and following up takes most of a working day each week for a modest batch, and if the founder is doing it that is the most expensive day the company has. Add a monthly fee for an address finder and a sending tool, and a few weeks of domain warmup before any volume is safe.

## Autonomous marketing

https://kamugrows.com/glossary/autonomous-marketing

Autonomous marketing is marketing in which software decides what to do next, does it and reports back, while a person sets the limits it works inside rather than approving each step.

Autonomous marketing is marketing where the software chooses the action. A rule engine executes a decision a person made earlier; an autonomous system makes the decision itself, from a plan and from what it can see right now, and then acts on it. The person's job moves from operating the tool to deciding what it may do and reading what it did.

## The ladder from automation to autonomy

The first rung is automation: if this, then that. A trial starts and five emails go out over two weeks. A lead visits the pricing page twice and gets a tag. Every branch was written by a person in advance and the tool takes no decisions of its own. It is reliable, and it is blind, since anything nobody anticipated does not happen.

The second rung is assistance. A model drafts and a person sends. The copy arrives faster and the judgment about whether to send it, and to whom, is still human. This is where nearly all marketing software with "AI" in the name sits today, and it is a real improvement on a blank page that changes nothing about who has to show up.

The third rung is autonomy. The system decides what to work on, does the work, and reports what it did and why. A person sets the limits in advance: how many emails a day, which channels, what budget, which calls need approval. Inside those limits it acts without asking, and at the edge of them it stops and puts the question to the person.

## What has to be true for it to be safe

Autonomy without boundaries is an unattended process with your domain reputation and your money at stake, so the boundaries matter more than the intelligence. Four of them do most of the work.

Hard caps live in code. A daily sending limit the model could talk itself past is a suggestion; one enforced by the sending function is a cap.

Every action is logged with the reason it was taken. Supervision during the work is impossible by definition, so the check happens afterwards, and that only works if each entry says why.

Refusals are built in: unsubscribes honoured permanently, addresses verified before sending, ad campaigns created paused, no invented testimonials, no tracking pixels. None of those is a judgment the system should be trusted to make well, so each is written as something it cannot do at all.

Irreversible decisions go to a person. Spend, a change of positioning, a new channel: the system writes up the case and puts it in a decision inbox, and nothing happens until someone approves or declines.

## Where it is today

Autonomy exists for bounded, repeatable work: finding people who match a profile, writing and sending one email, drafting an article for a phrase, posting what shipped, filing a listing. Positioning and pricing are still human, and much of what is sold under the name today is the second rung with the third rung's name on it. Kamu is one implementation of the third: it works every day on outreach, content, social, a landing page and listings from a plan it wrote, keeps the caps and refusals above in code, logs each action with its reason, and sends anything that spends money or changes the plan to the founder as a decision to approve.

Also called: autonomous marketing platform, agentic marketing, autonomous growth.

In practice:
- Automation: a trial signup triggers a fixed five-email sequence, identical for everyone, written once a year ago.
- Assistance: a model drafts three LinkedIn posts every Monday and a marketer picks one, edits it and schedules it.
- Autonomy: software reads the week's releases, decides a post is warranted, writes it, checks it against a banned-phrase list and publishes it within a daily cap, then logs why.

Q: Is autonomous marketing the same as marketing automation?
A: No. Marketing automation executes rules a person wrote in advance, so every action was decided at setup and the tool only repeats it. Autonomous marketing has the software make the decision itself, from a plan and the current situation, within limits set in advance. Automation cannot surprise you and cannot do anything nobody specified. Autonomy can do both, which is why the limits matter.

Q: What makes autonomous marketing safe to run?
A: Boundaries that live in code rather than in a prompt: hard daily caps, addresses verified before sending, suppression honoured permanently, campaigns created paused. A log that records the reason for every action, since nobody is watching while it works. And a rule that anything spending money or hard to undo goes to a person as a decision before it happens. Without those it is an unattended process with your domain at stake.

Q: What is the difference between autonomous marketing and agentic marketing?
A: Very little. Agentic marketing is the newer label and stresses that the software is an agent, meaning it takes actions towards a goal rather than answering prompts. Autonomous marketing stresses that it does so without a person in each step. Both describe the third rung of the same ladder, above automation and above a model that drafts for a person to send.

Q: Can autonomous marketing decide positioning or budget?
A: It should not, and a system that offers to is one to be careful with. Positioning depends on things no crawler can read, and a budget is the one number whose owner has to be a person. A well-built system will propose either, with its reasoning, and wait for approval. What it decides on its own is the bounded, reversible work underneath: who to write to today, which phrase to draft for, whether this week's shipping is worth a post.

## SaaS directories

https://kamugrows.com/glossary/saas-directories

SaaS directories are sites that list software products by category so buyers can find and compare them, giving a new product a backlink and a page that ranks for its own name.

SaaS directories are sites that list software products by category. A buyer uses one to find and compare options; a founder uses one to be found. The listing is a page on someone else's domain that describes the product, links to it, and often ranks for the product's name before its own site does.

## What a directory is for

A listing does three separate jobs. A backlink from an established domain is the durable one, and the reason to file even where nobody browses. Being somewhere buyers compare matters only on the review sites. A launch moment is on offer from exactly one directory below; the rest are listings.

## The ten worth a listing

The long tail of "submit your startup" sites is mostly link farms that do nothing for a real product and occasionally hurt. These ten have an audience or a domain worth having. Tagline and description are character limits.

| Directory | Good for | Tagline | Description | Watch out |
|---|---|---|---|---|
| Product Hunt | The biggest single-day traffic spike a new SaaS can get, and a durable backlink | 60 | 260 | One launch per six months; the tagline says what it does rather than why it is exciting |
| BetaList | Early adopters, for pre-launch or newly launched products | 60 | 300 | Rejects products long since generally available |
| Indie Hackers | A founder audience that converts for developer tools, plus a product page that ranks | 80 | 400 | Reacts badly to promotion; write it as a founder describing their own work |
| AlternativeTo | People already looking to replace a competitor; usually the best long-term traffic here | 100 | 500 | Must name the incumbents it replaces, honestly, or the community edits it out |
| SaaSHub | Comparison-driven listings and a solid backlink for little effort | 80 | 400 | None |
| G2 | Where B2B buyers check before buying; slow and review-gated | 80 | 600 | Worth nothing without real reviews; list once customers would write one |
| Capterra | Same buyer intent as G2, stronger in some categories; free listing, paid placement optional | 80 | 600 | None |
| Microlaunch | A small, genuine launch audience, far less crowded than Product Hunt | 60 | 300 | None |
| Uneed | Curated, with a real audience; free listings are queued, paid ones jump | 60 | 300 | None |
| There's An AI For That | High traffic, for genuinely AI-native products | 60 | 300 | AI products only; a non-AI product listed here reads as spam |

None of the ten has a public submission API, and nearly all need an account and a captcha, so a person still fills the form.

## Writing a listing, and what not to expect

Sixty characters is about eight words. The tagline that fits says what the product does and for whom, in the words a buyer would type. "Invoicing for freelance designers" fits and would pass on Product Hunt; "Reimagining how creative teams get paid" says nothing about what it does and gets rejected there. Write the 600-character version first and cut down from it, so the short versions are the same product in fewer words. Name competitors on AlternativeTo, where a buyer wants them, and nowhere else.

A listing is a slow asset: a backlink on day one, a trickle of visits over years, now and then a buyer who was comparing. It brings no traffic spike, with one exception. Product Hunt is a launch rather than a listing, it happens at most once per six months, and it should be planned for a day the product can hold the traffic rather than fired off when the form is found.

Filing listings is one of Kamu's jobs. It picks the directories worth the founder's time from this list, writes the copy inside each one's limits, and hands the founder the browser with the listing ready to paste.

Also called: software directories, startup directories, SaaS listing sites.

In practice:
- A product page on AlternativeTo that names the three incumbents it replaces and starts ranking for "[incumbent] alternative" within a few months.
- A Product Hunt launch planned for a Tuesday six weeks after the product could hold the traffic, with the 60-character tagline cut down from the long description.
- Not worth filing: a "top 500 startup directories" list where most entries are link farms with no audience, some of which can hurt the domain they link to.

Q: Are SaaS directories worth it for a new product?
A: Yes, for a specific and modest return: a backlink from an established domain, a page that ranks for the product's own name early on, and on the review sites a place buyers check before buying. They are not worth it as a traffic source, with the exception of a planned Product Hunt launch, and the long tail of submission sites is mostly link farms that give nothing back.

Q: How many SaaS directories should you submit to?
A: About ten, chosen for audience or domain rather than for count. The ones on this page cover a launch site, the early-adopter and founder communities, the alternative-search site, and the two review sites buyers check. Filing to a hundred takes days, brings nothing the first ten did not, and puts the product's link on sites that occasionally do harm.

Q: What gets a directory listing rejected?
A: Usually one of four things. A tagline that says why the product is exciting rather than what it does, which Product Hunt is strict about. A product that has been on general release too long for an early-stage site like BetaList. A comparison on AlternativeTo that does not name the incumbents or is not honest about them. And a non-AI product on an AI-only directory, which reads as spam and is treated that way.

Q: Can directory listings be submitted automatically?
A: Not today. Each of the ten directories here needs an account and clears a captcha, and none exposes a public submission API, so the most software can do is write the copy inside each directory's limits and hand a person the form with the text ready. A service that claims to file to hundreds of directories automatically is mostly filing to the link farms, since the ones with an audience do not allow it.

## Suppression list

https://kamugrows.com/glossary/suppression-list

A suppression list is the set of email addresses a sender must never email again, kept in one place so that no campaign and no tool can reach them by accident.

A suppression list is the set of addresses you have promised never to email again. It sits in front of every send, and the check is simple: if the address is on the list, the message does not go, whatever campaign it belongs to and whoever wrote it.

It exists because the law in most places requires an opt-out to be honoured for good, and because mailbox providers notice when people who asked to be left alone hear from you anyway. A complaint from one of them costs the sending domain more than the message could have earned. It is also the decent thing to do. Somebody said stop.

## What goes on it

Four kinds of address belong there. People who unsubscribed, by clicking the link or by replying with words to that effect. Addresses that hard bounced, because a mailbox that does not exist will not exist next month either. Anyone who marked a message as spam. And anyone who told you, in any channel, to stop writing. Once an address is on the list it stays there; a suppression list only grows.

## Suppressed is different from not interested

The mistake that causes real trouble is treating every no as the same event. A reply that says "we already use something" or "not right now" is a no to one pitch from one product. That conversation should close and that person should not hear about that product again, and nothing more follows from it. A reply that says "take me off your list" is consent withdrawn. It applies to the person, whatever you might sell them from now on, and it belongs on the suppression list.

The list also has to be one list. A company that runs three products with three outreach tools and keeps three unsubscribe lists has not honoured anything: the person who opted out of product A gets product B's sequence a month later, from the same company, and is right to complain. An address is suppressed for the whole sender, across every product and every person sending on its behalf, or it is not really suppressed.

## How Kamu applies it

When a reply asks to be taken off the list, or someone clicks the unsubscribe link, Kamu writes the address to a suppression list shared across every product in the account, and it checks that list before any send, before the address is even verified. A reply that says not interested closes that one conversation and touches nothing else: the person is not written to again about that product, and is not suppressed for the others.

Also called: Email suppression list, Do-not-contact list, Unsubscribe list.

In practice:
- A founder who replied "unsubscribe" to an email about product A, and never hears about product B from the same company either.
- An address that hard bounced in March and is still refused in September, even after a freshly imported list contains it again.
- Not on it: a prospect who replied "we already have a tool for this", whose thread closes without the address being suppressed.

Q: What is the difference between a suppression list and an unsubscribe list?
A: An unsubscribe list holds the people who clicked the link. A suppression list is broader: unsubscribes plus hard bounces, spam complaints and anyone who asked in words to be left alone. Tools use the two terms loosely, and the useful question to ask of any of them is whether the list is checked before every send from every product, or only inside the campaign it was built in.

Q: Does a suppression list expire?
A: No. A hard bounce could in theory be a mailbox that comes back, and a person who unsubscribed could change their mind, and the sender does not get to decide either of those. If someone wants to hear from you again they can write to you, and that new consent is theirs to give. Treat the list as permanent and only growing.

Q: Is a suppression list required by law?
A: The mechanism is not named in most laws; the outcome is. CAN-SPAM requires an opt-out to be honoured within ten business days and never sold or transferred. GDPR and PECR require that an objection to direct marketing is respected without conditions. A shared, permanent suppression list is the only practical way to meet those requirements once more than one person or tool is sending. None of this is legal advice.

Q: Should I suppress people who said not interested?
A: Not automatically. A no to one pitch is information about that pitch, and the person has not withdrawn consent to hear from you. Close the conversation, record the reason, and do not write to them about the same thing again. If they say stop, or ask to be removed, that is the moment the address goes on the suppression list.

## Email bounce rate

https://kamugrows.com/glossary/email-bounce-rate

Email bounce rate is the percentage of emails you sent that the receiving server refused or returned, worked out as bounces divided by emails sent.

Email bounce rate is the share of your sent email that came back. The formula is bounces divided by emails sent, times a hundred: send 200 emails and get 6 back, and your bounce rate is 3%. It is the first number a mailbox provider looks at when deciding what kind of sender you are, because it is the one number a sender cannot argue with. A message either reached a mailbox or it did not.

A high rate says something specific about you. It says you are writing to addresses you have not checked, which is what a bought or scraped list looks like from the receiving end, and providers respond by filtering the rest of your mail before anyone reads it. The bounces themselves cost nothing. The reputation they cost the sending domain takes weeks to recover.

## Hard bounce vs soft bounce

A hard bounce is permanent. The address does not exist, the domain has no mail server, or the receiving server has refused your mail outright. Nothing you do will get a message through, and sending again only repeats the damage, so a hard-bounced address goes straight onto the suppression list.

A soft bounce is temporary. The mailbox was full or the message was too large, the receiving server was busy or briefly unreachable, or greylisting asked the sender to try again later. Sending systems usually retry a soft bounce on their own for a day or two. An address that soft bounces repeatedly is usually a hard bounce that has not admitted it yet, and it is reasonable to stop after the second or third.

## What is a good bounce rate

The figure most often quoted is 2%, and it is a rule of thumb rather than a published threshold: mailbox providers do not say where their line is, and it moves with everything else they know about you. What can be said is that a verified list of people who actually work where you think they work bounces well under that, and that a rate climbing past a few percent is a list problem rather than bad luck.

Keeping it low is mostly done before the first send. Verify every address, so the ones that do not exist never leave the building. Stop after a hard bounce and never try that address again. Ramp a new domain slowly, because a bounce on a domain with no history counts for more than the same bounce on one with years of clean sending behind it. And treat a rising rate as a signal to stop and look at the list, since sending through it makes the next campaign worse.

## How Kamu handles bounces

Kamu checks an address before the first email to it, and an address that fails verification is marked as bounced without a message ever being sent. Delivery is recorded per email from the sending provider's own events, so "it never arrived" and "nobody replied" appear as different lines. A bounce breaker watches the last two weeks of sending: once at least twenty emails have gone out and 5% or more of them have bounced, sending for that product stops while the rate stays there.

Also called: Bounce rate, Hard bounce, Soft bounce.

In practice:
- 200 emails sent, 6 returned as undeliverable: a 3% bounce rate.
- A soft bounce from a full mailbox on Friday that delivers on the retry on Monday.
- A list exported from a CRM last touched two years ago, where many of the addresses belong to people who have since left.

Q: What is a good email bounce rate?
A: Under 2% is the figure most people repeat, and it is a rule of thumb rather than a number any mailbox provider has published. For cold email the honest target is as close to zero as verification can get you, because every hard bounce is an address that could have been checked in advance. A rate that is climbing matters more than any absolute level: it means the list is decaying or was never verified, and the fix is to stop and clean it rather than to keep sending.

Q: What causes a soft bounce?
A: A temporary condition at the receiving end: a full mailbox, a message over the size limit, a server that was down or busy, or greylisting that asks unfamiliar senders to retry. Sending systems usually retry on their own for a day or so. If the same address soft bounces on several attempts it is behaving like a hard bounce and should be treated as one.

Q: Do bounces hurt my sender reputation?
A: Yes, and faster than almost anything else. A hard bounce tells the receiving provider you are mailing addresses you did not check, which is what a purchased list looks like from their side. Providers weight it heavily because it is hard to fake and easy to avoid. A run of bounces on a young domain can get the rest of its mail filtered for weeks.

Q: Should I remove soft bounces from my list?
A: Not after one. A full mailbox is not a dead address, and most systems will retry and deliver. After two or three soft bounces to the same address across separate sends, treat it as hard: something at that mailbox is not going to change, and each further attempt adds to your rate without any chance of a reply.

## Email open rate

https://kamugrows.com/glossary/email-open-rate

Email open rate is the percentage of delivered emails that were recorded as opened, measured by a tracking pixel that loads when the message is displayed.

Email open rate is the share of delivered emails that registered as opened. It has been the headline number of email marketing for as long as there have been email tools, and it is still the first thing they show, which is a problem, because for several years now it has measured something other than people reading your email.

An open is counted by a tracking pixel. A one-pixel image with a unique address is placed in the HTML of each message, and when the recipient's mail client fetches that image, the sender's server logs an open. That is the whole mechanism, and everything wrong with the number follows from it.

## The formula

Open rate is unique opens divided by emails delivered, times a hundred. The bottom of the fraction is delivered mail rather than sent mail, so bounces come out first: a campaign that sent 1,000 emails, had 50 bounce and recorded 190 opens has an open rate of 20%, not 19%. Unique matters too. One person opening the same message four times is one open.

## Why open rates lie

The pixel only fires when images load. Clients that block remote images by default, which includes many corporate Outlook setups, record no open for a message somebody read closely. In the other direction, since Apple introduced Mail Privacy Protection in 2021, Apple Mail fetches every image in every message through Apple's servers whether or not anyone opens it, so anyone reading on Apple Mail with the feature on registers as having opened everything. Gmail's image proxy and the security scanners that fetch links and images do the same at smaller scale. The number now contains an unknown mix of people and machines, and no arithmetic separates them.

For cold email there is a second problem that has nothing to do with accuracy. The pixel is an image in HTML, and a first email to a stranger is supposed to look like something a person typed. A plain text message with no images is what a colleague sends; the same words wrapped in HTML with a hidden image are what a marketing platform sends, and filters have been trained on that difference. Tracking opens on cold email makes the email look like the thing it is trying not to be, and can cost you deliverability to learn a number you cannot trust.

What is worth reading instead is the chain that leads to a conversation. Delivery, confirmed by the receiving server, tells you the message arrived. A reply means somebody read it and cared enough to type; a positive reply, kept separate from the polite declines, means the targeting was right as well. If those are low, the fix is the same one that would have raised opens in the old days: write to people for whom the email is plainly relevant, use a subject line that says what the email is about, and send from a domain in good standing.

## Kamu does not track opens

Kamu sends plain text and puts no pixel in it, so it has no open rate to show. The reason is the one above: an invisible image is what turns a short personal email into something a spam filter recognises as marketing, and the number it would produce would not tell a founder anything a reply does not tell them better. What Kamu records instead is delivery for every send and every reply that comes back, sorted by what it says.

Also called: Open rate, Email opens.

In practice:
- 1,000 emails sent, 50 bounced, 190 unique opens recorded: a 20% open rate.
- A campaign to an audience reading on iPhones that shows opens on almost every message and two replies, where the opens were Apple's servers fetching the pixel.
- A recipient on a locked-down corporate Outlook who read the email twice and forwarded it, recorded as never opened.

Q: What is the email open rate formula?
A: Unique opens divided by emails delivered, multiplied by a hundred. Delivered is sent minus bounces, so bounced messages do not drag the rate down. Tools generally count one open per recipient however many times they return to the message, and they also count an open when the tracking pixel is fetched by a machine rather than a person, which is the reason to treat the result loosely.

Q: What is a good email open rate by industry?
A: The published tables vary by industry and by the vendor publishing them, and none of them mean much any more, because the pixel they depend on is fetched by Apple's servers and blocked by many corporate clients regardless of industry. Comparing your rate against a benchmark built from the same broken measurement tells you where your audience's devices sit more than how your email did. Compare reply rates instead, against your own previous sends.

Q: How do I improve email open rates?
A: Improve the things an open rate used to be a proxy for. Send to people with a specific reason to care, write a subject line that describes the email in plain words, send from a warmed domain with SPF, DKIM and DMARC in place, and keep bounces near zero. Then judge the result by replies rather than by opens, since the opens will move for reasons that have nothing to do with you.

Q: Does open tracking hurt deliverability?
A: It can. The pixel requires HTML, and a short message padded with markup and an invisible remote image scores differently with filters than the same words as plain text. Marketing platforms carry that cost because their newsletters are HTML anyway. For a cold email that is meant to read as a note from one person to another, it is a cost with no matching benefit.

## Email deliverability test

https://kamugrows.com/glossary/email-deliverability-test

An email deliverability test sends a message to controlled addresses and reports where it landed and what about the sender's setup or content could get it filtered.

An email deliverability test finds out where your email lands before your prospects do. You send a message to an address the testing service controls, or to seed mailboxes at Gmail, Outlook and Yahoo, and it reports what it saw: whether the message arrived, which folder it went to, and what about your setup or content a filter might hold against you.

The checks fall into four groups. Authentication: whether SPF, DKIM and DMARC records exist and pass for the message you sent. Reputation: whether the domain or IP is on a public blocklist such as Spamhaus. Content: HTML weight, broken links, a missing unsubscribe and the words older filters score against. Placement: for a seed-list test, which folder the message landed in at each provider.

The two kinds answer different questions. A one-off check, of the kind mail-tester.com or MXToolbox run, grades the technical setup around a single message. A seed-list placement test, which several paid services offer, sends the same message to dozens of real mailboxes across providers and reports what share reached the inbox. The first checks your plumbing; the second samples the outcome.

## How to test email deliverability

1. Send the real message: the account, the domain and the exact copy you plan to send, unsubscribe line included, because a filter grades the whole thing together.
2. Check authentication first. A failing SPF, DKIM or DMARC record explains most bad results on its own and is fixed in DNS rather than in the copy.
3. Look up the domain and IP on the public blocklists. A listing has to be cleared first; no change to the copy will get past it.
4. Read the content flags with some scepticism. Filters weigh sender history far above wording, so a "spammy word" warning is a low priority next to a missing record.
5. If you run a placement test, send to seeds at the providers your prospects use. Business mailboxes are usually Google Workspace or Microsoft 365, and a result for Yahoo says little about that audience.
6. Run it again when something changes: a new domain, a new sending tool, a DNS edit, or a sudden fall in replies. Otherwise once a month is plenty.
7. On a bad result, fix the records, get off any blocklist, then lower volume rather than pushing through. A domain that keeps sending into spam confirms the filter's judgement.

## What a test cannot tell you

A test sees a snapshot from mailboxes that are not your prospects'. Their filters have a history with you that a seed account does not: how that company's people treated your earlier mail, and what the rest of your domain has been sending for months. A clean result on day one of a new domain is the least informative there is, because nobody has any history to judge yet. A test cannot show reputation over time, which is what actually decides placement, and it cannot tell you whether anyone read the message. It tells you the plumbing is sound, which is necessary and is not the same as the message being welcome.

## What Kamu records instead

Kamu does not run placement tests. It records what the receiving server said about every email it actually sent, delivered, bounced, marked as spam or failed, from the sending provider's own events. That answers a different question, whether the mail you are really sending to the people you are really writing to is arriving, and it answers it continuously rather than once. A test before the first send is still worth doing; the two are not substitutes.

Also called: Deliverability check, Inbox placement test, Spam test.

In practice:
- A one-off check that returns a low score because DMARC is missing, fixed by adding one DNS record.
- A seed-list test showing the same message in the inbox at Gmail and in junk at Outlook, pointing at a reputation gap rather than a content problem.
- A perfect score on a fresh domain, followed a week later by real mail landing in spam once volume started.

Q: How do I test email deliverability for free?
A: Send a message to a free one-off checker such as mail-tester.com, and look up the sending domain and IP with MXToolbox. Between them you get the authentication results, blocklist status and the content flags. What neither gives you for free is a placement sample across many real mailboxes, which is the part the paid seed-list services charge for. For a small sender the free checks catch most of what goes wrong.

Q: How often should I run a deliverability test?
A: Whenever something about the sending setup changes, and otherwise roughly monthly. A new domain, a new tool, a DNS change, a new template or a sudden drop in replies are all reasons to test now. A domain sending steadily, with low bounces and replies coming in, is passing a live test every day, and the formal one adds little.

Q: Why does my test pass but my emails still land in spam?
A: Because the test sees a mailbox with no history of you and your prospects' mailboxes have one. Reputation is built per receiving provider from how its users treated your earlier mail, and a seed account has none of that. A passing test rules out the mechanical problems, which is useful. What remains is how you have been sending: volume on a young domain, bounces, complaints, and whether anyone replies.

Q: What is a good email deliverability test score?
A: The scores are each tool's own scale and are not comparable between tools. A score at the top of the scale means the mechanical checks came back clean, and that is the useful reading: what was flagged, rather than the number. Treat any failing authentication record or blocklist entry as something to fix regardless of the overall score.

## Cold email template

https://kamugrows.com/glossary/cold-email-template

A cold email template is a reusable structure for a first email to someone who has not heard from you: an opening observation, a reason it is relevant, a small ask and a plain sign-off.

A cold email template is a pattern for a first email to a stranger that you can reuse across many recipients. Templates are also the reason most cold email fails. A template is something many people have already received, and a recipient who has seen the shape before stops reading at the first familiar line. "I hope this finds you well" and "quick question" are templates announcing themselves.

The phrase has a second life: many people searching for it are job seekers writing to a hiring manager. This page is about business email, one company writing to a possible customer, though the shape works for both.

What transfers from one good cold email to the next is the shape. The words do not, because the words have to come from the person you are writing to.

## The shape of a cold email that gets replies

Open with a short, specific observation about the person: something they shipped, wrote, hired for or said, in one sentence, with no compliment attached. Follow it with one sentence on why that makes your product relevant to them now. Then make a small ask, something that can be answered in one line rather than a meeting. Sign off plainly, with your name and nothing else. Under a hundred words in all, and it should read like something you could say to the person over lunch.

## Subject lines

A cold email subject line has one job, which is to say what the email is about. Describe the email in three to six words, the way you would label a file. Lowercase is fine. Avoid anything that could pass for a newsletter or a promotion: manufactured urgency, or a personalisation token obviously filled in by software. "your onboarding checklist" beats "A quick idea to grow your business", because it tells the reader which email this is before they open it.

## An example

Suppose you sell a tool that turns a changelog into release notes customers can read, and a small SaaS company has just published its third changelog entry this month on a bare GitHub page, the latest covering a pricing change. Its founder is Anna.

*Subject: your changelog*

Hi Anna,

You have shipped three changelog entries this month, all on the GitHub page, and the last one covered a pricing change your customers probably wanted to hear about directly.

We turn a changelog like that into a release note your customers can read, in your voice, without anyone on your side writing it up.

Would it be useful if I sent you what your last three entries would look like?

Sam

Seventy-odd words, each sentence doing one job. A minute on their GitHub page produced the first line, and nothing in the email could be sent to anyone else.

## Cold email format

Under a hundred words, so it fits on a phone screen without scrolling. Plain text rather than HTML, with no images and no formatting beyond line breaks: a designed email is what a filter and a reader both recognise as marketing. One link at most, and only to your own product. A real unsubscribe line and your postal address at the bottom: the law requires them in most places, and they tell the reader they can make it stop.

Kamu writes each cold email from something the person actually did, such as a release they shipped or a job they posted, which is the opposite of a template: the shape above stays fixed and the words are new every time.

Also called: Cold email format, Cold email structure, Cold email example.

In practice:
- An email that opens with the pull request the recipient merged on Tuesday, and asks one question about it.
- A template that opens "I hope this email finds you well", recognised and deleted before the second line.
- A subject line reading "your pricing page" rather than "Quick question about growth at {company}".

Q: What is the best cold email template?
A: The one the recipient has never seen, which rules out anything you can download. Treat a template as a shape rather than as words: an observation about the person, a sentence on why it makes you relevant, a small ask and a plain sign-off, under a hundred words. Then write the words fresh for each person. A template worth reusing is one nobody could recognise as reused.

Q: How long should a cold email be?
A: Under a hundred words, and shorter is usually better. The email should fit on a phone screen without scrolling. Length is a signal in itself: a long first message tells the reader you are sending the same thing to many people and had to cover every angle. Three or four short paragraphs, one idea each, is enough to earn a reply, and a reply is the only goal a first email should have.

Q: What makes a good cold email subject line?
A: It describes the email in three to six plain words, so the reader knows which message this is before opening it. Lowercase is fine. It should not promise anything, invent urgency, or ask a question the body does not answer. The best test is whether the subject line would look normal on an email from a colleague: "your changelog" passes, "Boost your pipeline this quarter" does not.

Q: Should I use HTML or plain text for cold email?
A: Plain text. A first email to someone who does not know you should look like it was typed by a person, and HTML with images, buttons and a tracking pixel looks like it came from a marketing platform, to the reader and to the spam filter alike. Plain text also renders the same everywhere and cannot break. If you need a link, one is enough, and it should go to your own product.

# Contact

- Website: https://kamugrows.com
- Email: hello@kamugrows.com
- x.com: https://x.com/kamugrows
- instagram.com: https://www.instagram.com/kamugrows/