Hello everyone! I'm working on a bot for Microsoft Teams using Azure AI Foundry and Azure OpenAI. I've successfully set up an Azure AI Search index that pulls data from SharePoint, and I've been using the GPT-4o model with vector embeddings. In the Azure AI Foundry chat playground, everything runs smoothly and the bot provides the right responses based on the context from SharePoint.
When I test it locally in Visual Studio, it works perfectly while I'm debugging through Teams in Chrome. However, the trouble starts after I deploy it to my development Teams environment for publishing. The bot does not respond at all, and I've noticed the validation fails during the process. I used the m365 agent toolkit in Visual Studio to deploy, provision, and publish, but I suspect something might be amiss with my bot manifest, authentication, or Azure app registration.
Has anyone experienced similar issues transitioning from local development to a published Teams bot? I'd appreciate any insights or examples from your own setups!
2 Answers
I haven’t worked with those specific tools, but it's best to check the logs from your agent server to see if there are any requests hitting the server when you try to run the bot after deployment. If you can access those logs, share them here! If there are no logs, the issue might be with the integration itself, meaning it’s likely not receiving requests properly.
It sounds like you might need to double-check the settings in your bot manifest and ensure the authentication is correctly set up. When you deploy, make sure that the configuration for the Azure app and service principal matches what you have locally. Any discrepancies there could cause the bot to fail validation after publishing.
Related Questions
Biggest Problem With Suno AI Audio
How to Build a Custom GPT Journalist That Posts Directly to WordPress