Skip to main content
cancel
Showing results for 
Search instead for 
Did you mean: 

Get Fabric Certified for FREE during AI Skills Fest. This week only. Secure your voucher now.

Reply
z99
New Member

Fabric Data Agent in M365 Copilot

I created Fabric Data Agent and published it in M365 Copilot (picture 1)

z99_0-1776243607742.png

 

I tried to test it in M365 Copilot and succeeded, but I made changes in my Fabric Data Agent and republished it in M365 Copilot, then I made a new conversation and somehow I did not have permission to my Fabric Data Agent. I had already tried it using the M365 Copilot account which I also used to create fabric data agents. However, when I tried to continue the last conversation, it worked. Would this be considered an issue, or what should I do?

1 ACCEPTED SOLUTION
arabalca
Impactful Individual
Impactful Individual

Hi @z99 ,

What you're describing is an interesting behavior. I'm not certain about the exact cause, but based on how M365 generally works, here are a few hypotheses:

Possible cause 1 — Permission propagation

I understand that when you republish an agent, M365 Copilot might be generating a new internal reference to it. If that's the case, the associated permissions would need to propagate again across the tenant, which in other M365 services (SharePoint, Teams apps) can take anywhere from a few minutes to an hour. The old conversation would keep working because it still holds a cached reference to the previously authenticated version.

Possible cause 2 — Conversation state

I understand that M365 Copilot might be binding the agent context to the specific conversation where it was originally set up, meaning a new conversation tries to resolve the agent from scratch and at that moment something isn't ready yet.

Possible cause 3 — Incomplete publishing window

It could also be that after republishing there's a short window where the agent is technically published but not yet fully available for new sessions.

 

In conclusion, I don't think this is a permanent error or a configuration issue on your end. Everything points to a transient behavior related to how M365 handles agent availability after a republish. The most reasonable thing would be to wait a few minutes and try again in a new conversation. If the issue repeats consistently and doesn't resolve on its own, it would be worth reporting to Microsoft as unexpected behavior.

 

If this helps solve your problem, I’d appreciate it if you could mark it as the accepted solution and give it a like 👍

Thanks!

View solution in original post

1 REPLY 1
arabalca
Impactful Individual
Impactful Individual

Hi @z99 ,

What you're describing is an interesting behavior. I'm not certain about the exact cause, but based on how M365 generally works, here are a few hypotheses:

Possible cause 1 — Permission propagation

I understand that when you republish an agent, M365 Copilot might be generating a new internal reference to it. If that's the case, the associated permissions would need to propagate again across the tenant, which in other M365 services (SharePoint, Teams apps) can take anywhere from a few minutes to an hour. The old conversation would keep working because it still holds a cached reference to the previously authenticated version.

Possible cause 2 — Conversation state

I understand that M365 Copilot might be binding the agent context to the specific conversation where it was originally set up, meaning a new conversation tries to resolve the agent from scratch and at that moment something isn't ready yet.

Possible cause 3 — Incomplete publishing window

It could also be that after republishing there's a short window where the agent is technically published but not yet fully available for new sessions.

 

In conclusion, I don't think this is a permanent error or a configuration issue on your end. Everything points to a transient behavior related to how M365 handles agent availability after a republish. The most reasonable thing would be to wait a few minutes and try again in a new conversation. If the issue repeats consistently and doesn't resolve on its own, it would be worth reporting to Microsoft as unexpected behavior.

 

If this helps solve your problem, I’d appreciate it if you could mark it as the accepted solution and give it a like 👍

Thanks!

Helpful resources

Announcements
June Fabric Update Carousel

Fabric Monthly Update - June 2026

Check out the June 2026 Fabric update to learn about new features.

Fabric SQL PBI Data Days

Data Days 2026 coming soon!

Sign up to receive a private message when registration opens and key events begin.

New to Fabric survey Carousel

New to Fabric Survey

If you have recently started exploring Fabric, we'd love to hear how it's going. Your feedback can help with product improvements.

Top Solution Authors
Top Kudoed Authors