Forum Discussion
KanchanKumbhar
Jun 12, 2025Copper Contributor
Repeated Connection Prompts and "Send Failed. Retry" Errors When Using SharePoint Lists in Copilot
Hi everyone,
I'm working on integrating SharePoint lists as a knowledge source in a Copilot agent. The setup works initially, but I'm encountering persistent issues when testing the agent or accessing it via Microsoft Teams/SharePoint site after publishing.
Issue Details:
- Initial Connection Prompt: When I first test the agent or access it from Teams/Site, it prompts me to connect to the SharePoint data source. This is expected and works fine.
- Repeated Prompts: After the initial connection, the agent repeatedly asks me to reconnect to the SharePoint list, even within the same session. This happens multiple times during a single conversation.
- Error Message: Frequently, I receive the error message:
"Send failed. Retry"
This disrupts the conversation and makes the agent unreliable for end users. After 2-3 retry it worked - Expected Behavior: I expect the connection to persist once authenticated, and for the agent to access the SharePoint list seamlessly throughout the session.
What I’ve Tried:
- Verified that the SharePoint list is correctly configured and accessible.
- Ensured that the data source connection is added and published in the Copilot Studio.
- Tested in both the Copilot Studio test environment and Microsoft Teams and its working fine on Copilot Studio test environment.
Questions:
- Is this a known issue with SharePoint list integration in Copilot agents?
- Are there any best practices to ensure persistent authentication or session handling?
- Could this be related to Teams-specific permissions or token expiration?
- Any workarounds or configuration changes that could help resolve this?
Any help or guidance would be greatly appreciated!
Thanks in advance.
1 Reply
Sort By
- Martin-Apps4RentIron Contributor
Yes, you can store the full conversation between the user and the agent in a custom Dataverse field. You’ll need to use Power Automate to capture messages during the chat. Create a cloud flow that triggers on each message (user + bot), appends it to a variable, and finally writes that full text into your Lead > Conversation multiline text field. It’s not built-in by default, but this method works well.