search

crewAIInc / crewAI

Framework for orchestrating role-playing, autonomous AI agents. By fostering collaborative intelligence, CrewAI empowers agents to work together seamlessly, tackling complex tasks.
https://crewai.com
MIT License
19.37k stars 2.68k forks source link

Iteration on Agent with human feedback , if human does not approve it should not move forward #960

Open rshah1990 opened 2 months ago

rshah1990 commented 2 months ago

I am experimenting with a sample code that involves a team of two AI agents.

Agent 1 is responsible for crafting a travel blog,

Agent 2 then transforms this blog into a Facebook post, essentially creating marketing content.

My intention is to ensure that Agent 1 collaborates with a human user for revisions until the user approves the content. Only then should it be passed on to Agent 2 for further re-purposing. I initially believed that incorporating human feedback would facilitate this process. However, despite specifying this in the task description, the process advances to the next step without the desired iteration.

theCyberTech commented 2 months ago

How have you se up your tasks? did you enable as per the docs?

https://docs.crewai.com/how-to/Human-Input-on-Execution/

rshah1990 commented 2 months ago

Yes I did as per documentation, also I have specifically mentioned in the task description do not move forward without human approving it, it takes input modifies the output but move to next agent. It does not iterate on the same step.