Could we have a built-in tool for spawning additional chat completion calls?
This would permit the use of assistant API to build ad-hoc processing pipelines (for prompt refinement and testing, chain-of-thought, etc).
Might need a parameter to limit total calls or recursion depth.
Hi! Welcome to the forum!
While I personally don’t recommend it, a lot of people are starting out with tools like LangChain to accomplish this goal.