Been bootstrapping a tiny B2B SaaS for about a year now and runway anxiety is a constant low hum in the background. One thing that genuinely changed how I work is using LLMs as a sounding board before I commit to any feature or a new positioning angle.
Not in a promptengineery way. More like I dump my halfformed thinking into a chat, describe the problem I think customers have, and ask it to poke holes in my assumptions. It catches stuff I miss because I'm too close to it.
What I keep wondering is how much of that feedback is actually useful vs. the model just patternmatching on generic startup advice it was trained on. Sometimes the response feels sharp and specific. Other times it feels like it pulled from a mediocre blog post from 2019.
Curious if other founders or people building products are using AI in this prevalidation stage, or if most of the use is further downstream once the thing already exists. Also whether you trust the output at all for something this highstakes, or you treat it more like a rubber duck that occasionally says something smart.
The cost of being wrong when you're bootstrapped is pretty real, so I'm trying to figure out where AI actually earns its place vs. where I'm just using it to feel productive.
[link] [comments]