Power Platform Community Forum Thread Details

I am experiencing an issue with the topic cancellation functionality in our copilot studio deployed on Microsoft Teams. We have implemented a flow that involves 4-5 topics to loop over a list of PDFs, generate summaries using generative answers, and write the responses to an Excel sheet.

Current Implementation:

The virtual agent processes multiple topics to handle PDF summarization and updates an Excel sheet.
We tried implementing a cancellation flow by adding trigger phrases to stop the flow midway, if needed.

Expected Behavior: The virtual agent should immediately cancel the ongoing flow and stop execution upon receiving the cancellation command.

Current Behavior: The virtual agent does not cancel the flow immediately. It only shows the cancellation message after all topics have been executed.

Steps Taken:

Added trigger phrases for cancellation.
Published the bot to Microsoft Teams.

Questions:

Has anyone faced a similar issue with topic cancellation in copilot studio bot on Teams?
Are there any best practices or specific configurations needed to ensure immediate topic cancellation?
Could there be common pitfalls or mistakes that I might be overlooking?

Any insights or suggestions would be greatly appreciated. Thank you!

Categories:

Topic creation & management

Set Cancellation Topic as Highest Priority: Ensure that the cancellation topic has the highest priority in Copilot Studio, so it interrupts other topics as soon as the cancellation trigger is detected. This can help the bot respond to cancellation requests immediately, even in the middle of ongoing processes.
Use Interruption Handlers: Configure an interruption handler in Copilot Studio to actively listen for the cancellation command throughout the process. Set up cancellation checks within each topic, especially before initiating each new PDF summary or Excel write operation, allowing the bot to stop more promptly.
Divide Long Processes into Smaller Segments: Break down the PDF summarization and Excel update processes into smaller segments or batches. This way, the bot frequently pauses to check for cancellation commands, ensuring it doesn’t get “stuck” in a long-running process without listening for the cancellation input.