I have been on this so-called entrepreneurship path for the better of ~9 months now. Among other things I have:

• worked for free for a couple friends’ startup
• did contract work for a bit
• built a bunch toy prototypes
• spoke to many people across various industries to figure out startup ideas
• met people on various stages of the entrepreneurship journey (many of whom via the YC cofounder matching platform). Of these people:
  • One has actually gone on to be invested by YC
  • One has moved to SF to work for a YC startup
  • One has taken up a job as “Head of AI” at Airwallex

If I were to psychoanalyze myself, I have a firm high-level conviction in the potential of LLMs and the viability of doing a startup at this point in time, but for a long time, I am having trouble cashing out this conviction as concrete bet, to work on a concrete problem. It’s kinda funny listening to all the hot takes hyping up the potential of AI and how this is the best time to do a startup, even as I look around and struggle to make a concrete bet myself. This is despite being able to witness, say, AI-enabled text editors and ChatGPT becoming significantly better in the span of a few months, to say nothing of the ever increasing benchmark performance.

The conventional startup wisdom is to apply LLMs to B2B workflows, but a key problem I’ve faced is , if you’re building something to solve other people’s problems, the feedback loop from user interview → prototype → user testing is just too long. In contrast, if you are your own user, you can build for yourself and shorten the feedback loop and have conviction in your own taste.

I don’t think I can build a pure consumer startup, but I think I am working on a problem I face and I am betting that there are enough people who are like me who would like a solution too. Stay tuned.