Primary sources. Published arithmetic. No courses.
Pattern break · Edition 003
Somebody pulled 5,079 Stripe-verified projects off the TrustMRR API in March and ranked them by category. Marketplaces: $1,170 median MRR. Sales tools: $711. E-commerce: $599. AI: $156 — last, and below the $169 median of the whole set. Meanwhile the product doing $208,089 in verified revenue over the last 30 days is a social media scheduler whose own README calls it an alternative to Buffer. It did not become an AI product. It became a tool that AI can call.
Nevo David published a post in November 2024 saying his open-source scheduler made $700 a month. By the time he sat down with Mixergy in March he was at $45,000 and naming the cause precisely: not a model, not a feature, but a command-line interface, an MCP server and a skill listing on an agent hub. Stripe says he is now clearing two hundred thousand a month across 6,648 subscriptions.
The part worth stealing is what he stopped doing. The same company had already spent $12,000 on SEO and, in its own account, got minimal traffic for it. The channel that worked was free to publish and is still free to publish. That is the entire trade: in 2026 the cheapest distribution available is not a human audience, it is the tool list of every agent on a registry — and the people competing for it are all busy building AI products instead, which the data says is the worse half of the market.
TrustMRR verifies through the merchant's own Stripe key and refreshes hourly, so this is not a founder's screenshot. Postiz: $208,089 over the last 30 days, $225,757 total MRR, 6,648 active subscriptions, growing 15% month over month, ranked #89 on the board. Founded July 2024. The listing's own description of it: "the ultimate agentic social media scheduler tool."
"I am making $700 monthly with my open-source scheduling tool." The post is mostly a licensing argument — he advises against dual licensing and against holding SSO hostage, and points at Tailscale's seat-based model instead. There is no growth hack in it. What there is, is a dated, self-published floor, which is what makes the later verified figure legible as a curve rather than a claim.
At $45,000/month and climbing, David describes three artefacts: a CLI so an agent types postiz integration list rather than assembling a request body, an MCP server, and a skill published on an agent hub. The reason is token economics — every character in the payload is billed to whoever is running the agent, so the terse interface wins the integration. The inflection he names is not a campaign: an unknown developer wired up an agent called "Larry" that posted to TikTok through Postiz and wrote it up, and the writeup travelled.
Pulled from the TrustMRR API, so every row is Stripe-backed. Half of all projects earn under $169/month; the top quartile clears ~$800; the top decile clears $10,000. By category the medians run marketplace $1,170, sales tools $711, e-commerce $599, AI $156. By age: $148 in year zero, $656 in year two, $2,399 in year four. Only 48 projects — 0.9% — clear $10k/month while still growing 50% monthly.
Not MIT. AGPL-3.0 obliges anyone who runs a modified copy as a network service to publish their modifications, which is why handing out 6,500 forks of a hosting business does not end the hosting business. The banner at the top of the README is not a feature list: "NEW: check out Postiz agent CLI! perfect for OpenClaw and other agents." The product's stated identity, in the repository description, is "the ultimate agentic social media scheduling tool."
Paginated the public registry API, deduplicated to latest version per name: 101,219 entries collapse to 31,309 distinct servers, so the "100k servers" figure in circulation overstates by about 3.2×. 22.6% ship no source repository at all. The number that dates the opportunity is the registration rate — 14.6× year over year, four days ago. The shelf is free because it is new.
Postiz bills $208,089 across 6,648 subscriptions, so average revenue per subscription is $31.30 — barely above the $29 entry tier, meaning almost the entire book sits on the cheapest plan. Run that backwards and $5,000/month is 173 entry subscriptions, or 2.6% of the subscriber count Postiz built in twenty-six months. The capital side is smaller still: a $6 DigitalOcean droplet plus a database and a cache instance is roughly $24/month, so $500 buys about 20 months of runway. The registry listing, the licence and the hub skill cost nothing.
Seven rungs. The expensive decisions here are all made before any code ships, and every one of them is a decision to not do the thing the market is currently doing.
The AI cohort has the lowest median revenue of the four categories measured across 5,079 Stripe-verified projects — $156/month against $1,170 for marketplaces. Crowding shows up in the median long before it shows up in the discourse. Build in a category with a boring, well-understood job to be done and a clean API surface.
AI $156 · e-comm $599 · sales $711 · marketplace $1,170The $12,000 Postiz put into SEO returned, by its own account, minimal traffic, and was discontinued. This is the single clearest use of the $500 in the whole route: not spending it. A budget that size cannot buy attention in 2026, and the channel that eventually worked did not charge for placement.
$12,000 spent → discontinued · $0 spent → 15%/moOpen source is not the giveaway it reads as. AGPL obliges anyone offering a modified version over a network to publish their changes, so the 6,500 forks are contributors rather than competitors. A permissive licence at the same star count would have funded someone else's hosting business. The licence, not the code, is what is defensible.
An agent pays for every token it emits. postiz integration list is a handful of tokens; the equivalent JSON request body is hundreds, billed to the operator on every call. The terse interface wins on cost, and cost is the only preference an agent has. This is a genuinely new design constraint and almost nobody is designing for it yet.
Listing an MCP server costs nothing and puts the product in the tool list of every agent that reads that registry. 22.6% of the 31,309 servers there ship no source repository, which means a listing that links a real, starred, auditable repo is in the better quartile of the shelf on day one, for free.
The verified ARPU of the reference business is $31.30 on a $29 entry plan — the upsell is doing almost nothing, and the forecast is therefore a count of accounts, not a blended guess. Five thousand a month is 173 of them. Anything that does not add accounts at $29 is not a growth activity.
$208,089 ÷ 6,648 = $31.30 · $5,000 ÷ $29 = 173Registrations went from 456 a month to 6,673 in a year. Free distribution channels are free during the interval between a standard existing and the incumbents noticing, and that interval is measured in quarters. Treat the registry the way an early advertiser treated cheap inventory: use it hard now, and expect the price to be set by someone else by the time it is obvious.
456 → 6,673 new servers/mo · 14.6× YoYPost-ready. Counts are computed live from the text and flag anything over 280.
🧵 Postiz is a social media scheduler. Its own README calls it an alternative to Buffer. Stripe-verified revenue, last 30 days: $208,089. In November 2024 the founder posted that he was making $700 a month. He never turned it into an AI product.
Someone pulled 5,079 Stripe-verified startups off the TrustMRR API and ranked them by category median MRR. Marketplace: $1,170 Sales tools: $711 E-commerce: $599 AI: $156 AI is last. The median across all 5,079 is $169/month.
Consensus in 2026: build an AI product. The only Stripe-verified revenue dataset says the AI cohort earns less than any other category measured. The trade that worked was the inverse. Build the boring tool. Make agents able to call it.
What he actually shipped, per his March 2026 interview: - a CLI, so an agent types "postiz integration list" instead of building a JSON body - an MCP server - a skill on an agent hub Every token is billed to whoever runs the agent. The terse interface wins.
Before that, the same company spent $12,000 on SEO and got, in its own telling, minimal traffic. Discontinued. The channel that worked cost $0 to publish. Registry listing: free. GitHub: free. The $500 is best spent by not spending it here.
The repo is AGPL-3.0, not MIT. 34,700 stars, 6,500 forks. AGPL forces anyone hosting a modified copy to publish their changes. That's why giving the entire product away doesn't cost him the hosting business. The licence is the moat. Not the code.
The numbers that make $5,000/mo reachable: $208,089 ÷ 6,648 subscriptions = $31.30 average. Entry plan is $29. So $5,000/month = 173 people paying $29. A $6 droplet plus a database and cache runs ~$24/mo. $500 = 20 months of runway.
The catch, counted four days ago: new MCP server registrations went from 456/month to 6,673/month year over year. 31,309 servers live. The agent shelf is free because it's new, not because it's generous. That interval is the whole opportunity.