I was wondering, if there is a way to simulate the functionality of AutoGen / CrewAI / chatDev, etc on Make, by creating a workflow that allows bi-directional, multi-agent chat to incorporate self-reflection?
eg. Agent A is a writer, and Agent B is an Editor. Agent A writes a draft, then Agent B reviews it, and provides feedback to Agent A, to have it improve the draft based on its critique. Agent A rewrites the draft based on the feedback. This back and forth between the two Agents iterates until Agent B is satisfied with the output, which is then ready to be presented.
Currently there doesn’t seem to be a way to do this in a simple manner. If I manually do:
Agent A → Agent B → Agent A,
for one iteration, and pass the same Thread ID to both, Agent A has already lost context of its previous call. Possibly because each Assistants module call results in a new Run ID in the background.
Has anybody tried to implement this? Could there be a way to do this in the current Make functionality? If not, is there an appetite to implement this in the future?