Have AI Agents Assign Tasks to One Another - Multi-Agents
under review
mohitlavatre20
When I perform tasks I want custom agent A to call custom agent B. Is this something possible?
Log In
John at Taskade
under review
Thanks again for all your feedback! We'd love to hear any comments or use cases you have, so please share them right here in this post. Stay tuned for more updates from us.
Don't forget to check out our latest changelog for all the new features and improvements: https://www.taskade.com/blog/updates/.
devinlewis
John at Taskade AI agent be able to create due dates or timers in the block.
Narek Zograbian
devinlewis: This is most likely a separate request. How would you like to see this implemented?
Narek Zograbian
Merged in a post:
Multiple custom agent background commands: more documentation or tutorial
smian
Hi there, is there a detailed documentation on how the background "Autonomous Multi-Agents work"? I'm following the docs here: https://help.taskade.com/hc/en-us/articles/22256943362323#8----deploy-multiple-gpt-agents, but I cannot seem to make multiple background commands work reliably. For example, if a custom agent has two background commands each with a separate role and task list, say one is a researcher and the other is an editor, will they have a dialogue with each other and determine the best combination of tasks the final goal? If multiple commands are enabled to run in the background, will they work together? Any additional clarity will be very much appreciated on how to leverage this potentially awesome feature. Thank you.
Narek Zograbian
Hey smian! It sounds like your feature request is similar to this one:
smian
Narek Zograbian sure, pls close this as duplicate.
r
ryantaskade
Hi there,
Currently, it is possible due to the Run agent command being released as an automation action. However, it seems there might be a discrepancy between the description and title of the post. Is this feature request focused on Agents having access to add-ons instead?
r
ryantaskade
Hi Smian,
That section of the article touches on how multiple commands can be run at the same time, do note that these runs are separate instances from each other.
No, the agents won't have a dialogue with each other. Currently, we do have some issues with our background command where the user's prompt may be overwritten by the system prompt, and we are actively working on addressing that. However, if you want to achieve responses from multiple agents, you can set up an automation flow where one agent's response is passed into another agent for further analysis.
More information on getting started with automation can be found here: