Has anyone tried to build a deep research agent using controlflow? Would love to learn how to achieve that kind of orchestration.

Yea I did a Marvin based workflow , I used the reasoning workflow as a base and it’s 5-6 stages and agents process is something like: give the task: 1: agent 1 constructs search queries 2: iterate through the list of queries , agent 2 collects urls from results 3: agent 3 scrapes urls 4: agent 4 extracts and ingests into RAG, 5 agent 5: does the same reasoning loop as the reasoning example in cf-> it queries the RAG and validates that the results it receives meets the criteria to accomplish the objective sufficiently, if not it starts over.. and does 1-5 again until it determines it can answer the objective satisfactorily