Is Heavy Use of result_as_answer=True a Bad Practice for Agentic Flows?

Hello, everyone! How are you doing?
I have a question and would appreciate your help.

Does the excessive use of the result_as_answer=True parameter in my flows risk making them less agentic and more like a programmed routine?

In my crews, I use result_as_answer quite a lot, mainly because the responses tend to be very large, which increases token costs.

Is this frequent use considered a bad practice? I would like to better understand whether my flow can still be considered agentic.

Thank you in advance!

Depends with your use case. If you are using the agents in a flow for agentic RAG purposes or in certain steps then you are doing fine. If it’s all you are doing without using the LLM powering the agent to use the tool results then you might not really need to use agents. With everything none trivial the correct answer is…it depends :sweat_smile: