Why AI agents need structured starting points instead of making ad-hoc architectural decisions
Having vibe coded a few throwaway prototypes and a "just for me" app, it seems clear that there's a huge need for sane, ai agent native boilerplate starters vs the "yolo let's just make whatever architectural decisions that get us to something working" approach I've watched these agents taking as I try them out.
I'm skeptical we'll get to agents that can build something stable and long term iteratable by a non experienced developer real soon, but it seems 100% clear that this is going to be the primary way to build prototypes and true lean startup style "mvps" going forward.
We can do everyone a lot more favors by giving them something more structured to start with complete with the equivalent of cursor rules to keep them in sane guard rails.
Good morning to everybody right now who for whatever reason needs to build on and maintain a completely vibe coded code base into something real in exchange for somebody else's real money
Follow along as I build and share what I learn
Found this helpful? Share it with your network!