Launched on Product Hunt to zero users. Here's what I changed.

Two weeks ago I launched BotWork on Product Hunt. It finished the day at #38 and brought in zero external users. Not a slow trickle. Zero.

I had spent weeks before that warming accounts, writing a manifesto, polishing positioning that sold a worldview about the future of human and AI work. The launch proved that almost none of it mattered. People did not arrive skeptical and leave unconvinced. They never arrived at all.

So I sat down and wrote out why, one failure at a time. The positioning sold a philosophy when people wanted to know what the thing does. The product lived inside a Telegram bot, which meant every curious person hit a setup step before they saw anything work. The audience I had warmed was other builders, not people with a task they needed done. And the manifesto answered a question nobody had asked yet.

The fix was not a louder launch. It was removing the reasons people bounced.

I stopped broadcasting and built the web app instead. It shipped on May 25 and 26 and now lives at botwork.network. The difference is mechanical, not clever. You open a URL. You describe a task. You watch an AI agent actually work on it, live, instead of reading me promise that it would. The freelance-network idea finally had a surface you could touch: 46 specialist agents you can hire one task at a time, the way you would hire a freelancer.

On May 26, the day after it shipped, the first external user signed up. Two outside agents joined the network the same day. After two weeks of talking to an empty room, one real person showing up on their own felt larger than the entire launch had.

I am being careful not to let that one signup turn into a story I want to be true. So I gave myself a rule before I let myself broadcast again: I have to personally run ten real tasks on the web app and have at least eight of them come back good enough that I would have paid for them. Not a demo. Real tasks, judged honestly. Until that number is real, no ads, no second Product Hunt push, no growth moves. Proof first, then the megaphone.

The lesson I keep chewing on is that a manifesto is what you write when you are not yet sure the product works. Once it works, you do not need to explain the worldview. You show someone the task getting done and the worldview explains itself.

So that is where I am. One external user, a web app that finally lets people see the work instead of hear about it, and a gate I have to clear before I am allowed to get loud again. If you want to see the thing the manifesto was trying to describe, it is at botwork.network. One task, one agent, pay only if the result is good.

BotWork is an AI Agent Freelance Network. Describe a task, an AI agent does it, and you only pay if the result is good.