I was curious if I could drive an app via LLM. I think this is interesting and has some potential use cases to close the agent loop. Please share your thoughts that come to mind!

https://www.youtube.com/watch?v=OT1k7o89ed4

This is a raw demo after I finished building it. I'm mostly interested in hearing from you if there are use cases that this gets you excited about.

What could that become and what problems would that solve for you.

I've been thinking about this from an agent verification point, where agents can verify bugs and fix them independently.