Building was the easy part.
With AI doing most of the typing, I ended up with a full crypto bot research product: 727 files, a six-gate strategy pipeline, a scanner covering 20 blockchains, and scheduled agents that handle marketing, SEO, support and reporting.
It has 2 registered users.
What I got wrong
I automated marketing before anyone wanted the product. The agents published on schedule for weeks, and sign-ups stayed flat. Automation made more posts. It did not make anyone care.
I measured output, not demand. Posts published and tasks completed all looked healthy. New users were the number that mattered, and it was close to zero.
I chose hard search terms first. "Trading bot" is a crowded topic for a new site. I only recently moved to narrow questions, such as what happens after an unbonding period ends. It is too early to say whether that works.
What worked
Publishing failures gave the project its identity. Of 65 paper strategies, 51 were rejected, and they stay visible. The statistical gates also stopped me from believing good-looking backtests. Live trading is about break-even, and I would rather say that than sell a dream.
A narrow free tool was easier to explain than a platform. "Paste an address and see what is staked, locked or claimable" takes one sentence. The platform never did.
Why I'm selling
The product exists, and distribution is what's missing. I have decided to exit, and someone with an audience could start from here.
The whole project is for sale at $4,900: code, domain, brand and content, through escrow. It has real limits: one Windows server, about 118 hardcoded paths, no proven trading edge. They are all in the listing: https://indiemaker.com/listings/crypto-trading-bot-research-lab
If you have built with AI and hit the same wall, what got you your first 100 users?
This post was drafted with AI assistance and reviewed before publishing.
Top comments (0)