What 10 HN comments taught me about shipping AI tools solo
I posted 10 comments on Hacker News over the past two weeks while building an open-source Chrome extension. The replies taught me more than any feedback form has.
A few notes I’m keeping:
The “shipping fast vs shipping right” debate is mostly a trap. Some people ship rough and iterate. Some polish longer and ship later. Both can work. The version that fails is the one that never leaves your machine.
Local models still get developers excited, but speed changes the workflow. At 6 to 11 tokens per second, a 200-line coding task can turn into a four-minute wait per turn. That is fine for planning or batch work. It is painful when you are debugging back and forth.
The “no strangers will pay” problem is rarely just about the product. Sometimes people try it once and do not come back. Sometimes the distribution channel is wrong. Sometimes there is a trust gap that SEO cannot close.
I am building AI Buddy around those lessons: small surface area, open-source code, no backend in the middle. It is still early, but the public replies are already shaping the product more than my private roadmap did.
Chrome Web Store: https://chromewebstore.google.com/detail/eigpaeoigklelmfgnkljhbjjbpohenpn
GitHub: https://github.com/mnbqwe10/ai_buddy
What did you learn from your first month of building in public?
One thing I've noticed after these kinds of posts is how easy it is to come away with a lesson when there are still a few competing explanations on the table.
A product not getting traction can teach you something real, but it doesn't always teach you the thing you think it did.
That was the part I found myself thinking about while reading this.