From what I've seen, you have to answer x amount of questions, based on your original request to avoid ambiguity. Your responses also matter, it's not simply based on 4 questions = being able to "go live", if it still lacks context after the 4th prompt, it will explicitly tell you that it needs a better idea of what's necessary to actually be built.
I do agree though, after 3 generally basic natural language prompts, I wouldn't trust it enough to handle my finances.
That being said, there are also legal implications that I'm sure this project would have to deal with before gaining any type of userbase.
I really enjoy this idea. I haven't taken the time to create an account yet, but was able to play a bit with the concept. What exactly is the selling point? In terms of architecture, is it just a wrap? How does it differ from asking claude, for example, to trigger trading actions, or to build a trading bot?
Otherwise, great concept, I appreciate the fact that you have to answer set amount of questions before anything is actually "built" to avoid a mess.
The audit trail question is interesting and I haven't seen it come up much. When an agent clicks through an ERP or edits a file, you've got logs, but how do you explain the "why" behind each decision to, say, a compliance team?
Curious if that's something you're thinking about or if it's too early.
That’s a great point, will work on trying to score not just the repo but author use too.