Turns out being the chatbot is a lot more exhausting than being the user. I definitely have a lot more sympathy for every "Could you clarify?" response now.
I’m especially curious about the “self-learning loop” — in practice, does it actually improve outcomes over time, or does it tend to reinforce suboptimal patterns?
And How much autonomy do the agents actually have in practice?
I’ve found that fully autonomous loops tend to need a lot of guardrails to stay useful.
Really like the concept. Could be interesting to map different branches or authors to instruments — might make the structure of the history more audible.
This looks very relevant for onboarding. Have you tried this with new engineers on a real project? I’d be curious whether it actually changes how quickly people build a mental model of the system.