AI Foundry Agent Publish to 365 and Teams as a Bot Service

Question

AI Foundry Agent Publish to 365 and Teams as a Bot Service

RSKenzz 96

We have an Azure AI agent in the new Foundry experience. Works all fine. Deployed it as a bot service to 365 and Teams via the Publish button and when testing it out in Teams or in the Bot web chat, I get the following -

Something went wrong, please retry. If issue persists, please use following identifiers in any support request: ConversationId = IwdimIyY03SIOXXKurZPY2-fr, activityId = IwdimIyY03SIOXXKurZPY2-fr|0000000

I have checked the messaging endpoint on the Bot service and I'm getting this -

{"error":{"code":"401","message":"Access denied due to invalid subscription key or wrong API endpoint. Make sure to provide a valid key for an active subscription and use a correct regional API endpoint for your resource."}}

Is anyone able to help please?

Brendan Hart 20 Reputation points

2025-12-10T22:43:49.1366667+00:00

We tested our activity protocol endpoint that was configured for the Bot and along with all the other errors above, we get this one. We've also added the proper RBAC roles to every resource, including the agent's identity and identity blueprint that foundry creates to being an Azure AI User, Search Index Reader, and Azure AI Project Manager at the resource group level (just to be sure). Seems like there is a bug on Microsoft's end and removing the knowledge store is not an option!
Peter Selberg 10 Reputation points

2025-12-29T14:47:37.42+00:00

Has there been any update on this issue? We are having the same problem where the agent works fine in Teams as long as there is no knowledge source, but with a knowledge source added it just gives vague errors.
Brendan Hart 20 Reputation points

2025-12-29T16:54:14.92+00:00

@Peter Selberg I have not heard anything from Microsoft or seen any updates on the forum. I am still encountering the same issue also

3 answers

Your answer

Brendan Hart 20 Reputation points

2025-12-10T22:43:49.1366667+00:00

We tested our activity protocol endpoint that was configured for the Bot and along with all the other errors above, we get this one. We've also added the proper RBAC roles to every resource, including the agent's identity and identity blueprint that foundry creates to being an Azure AI User, Search Index Reader, and Azure AI Project Manager at the resource group level (just to be sure). Seems like there is a bug on Microsoft's end and removing the knowledge store is not an option!
Peter Selberg 10 Reputation points

2025-12-29T14:47:37.42+00:00

Has there been any update on this issue? We are having the same problem where the agent works fine in Teams as long as there is no knowledge source, but with a knowledge source added it just gives vague errors.
Brendan Hart 20 Reputation points

2025-12-29T16:54:14.92+00:00

@Peter Selberg I have not heard anything from Microsoft or seen any updates on the forum. I am still encountering the same issue also

Answer 1

Szymon Balinski 5

I’m encountering exactly the same issue described in this thread.

I have an agent built in AI Foundry, connected to a Knowledge Base. In the AI Foundry playground everything works perfectly, the agent responds as expected and retrieves information from the knowledgebase without any issues.

The problem starts only after publishing.

When interacting with the deployed bot, I receive the same error reported here.

Something went wrong, please retry. If issue persists, please use following identifiers in any support request: ConversationId = 2GDclp5wMoqwzOUKl2ZT6-eu, activityId = 2GDclp5wMoqwzOUKl2ZT6-eu|0000000

It’s not limited to the Teams or Copilot deployment either, the issue appears even earlier, when opening the bot instance in Azure and using the "Test in Web Chat" option. The bot was created using the "Publish to Teams and Microsoft 365 Copilot" wizard directly from the agent in AI Foundry.

web-chat-problem

I’ve also assigned all required roles, including Azure AI User for the AI Foundry project, exactly as documented.

The suggested solution by Anshika Varshney "Removing it was a good approach" is not really a solution… considering that vector search is the whole point of RAG.

RSKenzz 96 Reputation points

2025-12-09T10:38:38.42+00:00

Did you try what I mentioned above, see below.
Nevermind found it, it's because I had a vector store attached to my agent. Deleting that fixed it.
Szymon Balinski 5 Reputation points

2025-12-09T10:48:25.8233333+00:00

In my case, though, I’d really prefer to actually fix the underlying issue rather than work around it by removing the vector store, especially since that’s a core part of the agent’s functionality. Hopefully we can get a proper resolution from Microsoft.
RSKenzz 96 Reputation points

2025-12-09T10:53:42.24+00:00

Oh yeah same, fingers crossed a fix is coming soon, but this is what I had to do in the interim.
Brendan Hart 20 Reputation points

2025-12-10T22:40:23.5066667+00:00

Agreed, removing the vector store is not an option! That's the whole point of an agent. Has anyone found a fix?
Anshika Varshney 15,625 Reputation points Microsoft External Staff Moderator

2025-12-29T15:34:46.04+00:00

Thank you for bringing this to our attention. I have escalated your issue to internal team. They are currently checking the details and working on resolving the issue. We’ll update you as soon as we have more information.
Manas R Mohanty 17,270 Reputation points Moderator

2026-01-08T06:04:24.8566667+00:00

Hi RSKenzz,

I am facing authentication issue while testing in webchat.

Issue persists even after adding project managed identity as Azure AI user or Account owner.

Syncing here internal Product group channels for the same

Could you help with subscription, email and other details to create support ticket and escalate to product group for tracking this issue in time.

Thank you for your inputs.
Manas R Mohanty 17,270 Reputation points Moderator

2026-01-09T07:59:01.6633333+00:00

Hi RSKenzz and all

I was able to get pass the authentication issue today and finally getting results after couple of."Please wait while I process your request" from Agent in Bot service.

It took some time to reflect the permission probably

I have used "Grounding with Bing Search tool" along "websearch"/ couple of files uploaded.

My configuration

Region - East US 2

Project created under non-Hub based Foundry resource.

Follow up queries

Could you confirm the issue in non-hub-based Foundry project too.

Could you share the region and presence of any virtual network

Looking forward to your inputs.

Thank you.
Deleted

This comment has been deleted due to a violation of our Code of Conduct. The comment was manually reported or identified through automated detection before action was taken. Please refer to our Code of Conduct for more information.
Szymon Balinski 5 Reputation points

2026-01-13T11:03:56.7466667+00:00
Hi @Manas R Mohanty ,

I have verified my configuration as per your suggestions, but unfortunately, it is still not working when connected to Azure AI Search.

Here is the updated status of my environment:

I am using a non-hub-based AI Foundry resource in East US 2, just as you described.

Knowledge Base - The agent is connected to Azure AI Search via Foundry IQ.

I have confirmed that the project has been granted the "Azure AI User" role.

Everything continues to work perfectly within the Foundry Playground.

The issue only appears after publishing and interacting through the Bot service.

The only thing that has changed is the error message. I am now seeing a 400 (Bad Request) instead of the previous authentication errors:

Encountered error=Response status code does not indicate success: 400 (Bad Request).. If issue persists, please use following identifiers in any support request: ConversationId = 30vrnIf8GfVeCSLP3Kmji-eu, activityId = 30vrnIf8GfVeCSLP3Kmji-eu|0000000

It seems that while the "Grounding with Bing Search" might work in your case, there is still a significant issue with the Azure AI Search / Knowledge Base integration when accessed via the Bot protocol.
Brendan Hart 20 Reputation points

2026-01-13T17:40:43.3833333+00:00

Hi @Manas Mohanty

I added the foundry resource's managed identity as an Azure AI Owner on the Agent Bot and I am also still getting the same error with the Azure AI Search / Knowledge Base integration. Per other posts on the forum and Github (see the discussion here: https://github.com/microsoft-foundry/new-foundry-portal/issues/71) this is also happening with other MCP Servers as I believe Azure Foundry treats a knowledge base and MCP Servers as the same when providing context for an agent.

Please let us know when this is resolved and if we are doing what you expected us developers to do in terms of resource permissions
Loset, Christian 5 Reputation points

2026-01-22T06:36:50.83+00:00

Is there any update on this one? The publishing to 365 and Teams is still not working, with no workaround?

Answer 2

RSKenzz 96

Nevermind found it, it's because I had a vector store attached to my agent. Deleting that fixed it.

0 comments

Answer 3

Hi RSKenzz,

Thank you for reaching out on the Microsoft Q&A.

Publishing an AI Foundry agent as a bot in Microsoft 365 and Teams involves a few key steps:

Create and configure the agent in AI Foundry Ensure your agent is fully tested in the Foundry playground and has the correct permissions for the intended use case.

Register the bot in Azure Bot Service This gives you an App ID and secret that Teams will use for authentication.

Enable Microsoft Teams channel In the Bot Service settings, add the Teams channel. This allows your bot to appear in Teams and interact with users.

Connect to Microsoft 365 If your bot needs access to Microsoft 365 data (e.g., calendar, emails), configure Microsoft Graph permissions in Azure AD and grant admin consent.

Deploy and test Package your agent logic behind the bot endpoint (via Azure Functions or App Service) and deploy. Then, test in Teams using the Bot Framework Emulator or by sideloading the app.

Reference docs:

Please let me know if there are any remaining questions or additional details, I can help with, I’ll be glad to provide further clarification or guidance.

Thankyou!

Anshika Varshney 15,625 Reputation points Microsoft External Staff Moderator

2025-12-03T10:18:27.95+00:00

Hi **RSKenzz,
**Glad to hear you’ve resolved the issue! Thanks for sharing the root cause having a vector store attached to the agent can indeed lead to unexpected behavior. Removing it was a good approach.

I’ve converted my earlier comment into an answer, could you please take a moment to mark it as Accepted? This helps others in the community with the same question find the solution more easily.