Hi Community,
We are encountering 2 inconsistent runtime errors while testing an agent in Copilot Studio's Test pane, which is preventing us from reliably executing and validating the conversation flow.
The primary errors being observed are:
- TopicNotFound: "The Topic with Id ca_agent.topic.selectthereleaserequesttype was not found in the definition. Please check that the Topic is present and that the Id is correct. Error code: TopicNotFound. Conversation ID: e69488f4-cb78-49c0-8a5f-fa8a9919b156"
- SystemError: "Something unexpected happened. We're looking into it. Error code: SystemError. Conversation ID: c445ca84-e10e-48ad-a1d5-ca96ba8060bb. Time (UTC): 7/14/2026 8:06:48 AM."
The agent flow is alternating between the 2 errors, at different steps of the guided conversation flow, for different studio topics.
Initially, TopicNotFound error was encountered when initiating the conversation flow through a user-selectable option set presented by the agent. The error references a topic that does exist in the agent; rather the error is being raised from within a node of that topic after the topic has already been successfully triggered and executed up to a certain point. To note, the error is not being raise by the agent's 'on error' default system topic.
However, the behavior has been inconsistent:
- The TopicNotFound error was initially observed for a specific topic, but after some time it stopped occurring for that topic without any changes being made.
- The same error subsequently began appearing for other topics in the flow despite no configuration changes being made to any of them.
- The issue was also observed when attempting to trigger the topic using direct trigger phrases associated with them.
- The errors do not appear to be consistently tied to a specific topic or step.
In addition, SystemError is also being encountered during agent execution. Similar to the TopicNotFound error, it is occurring across different points in the flow and is preventing reliable testing of the conversation.
As part of the investigation:
- Topic IDs were reviewed in code view.
- Topic references and navigation paths were reviewed.
- Referenced topics were verified to exist.
- Copilot Studio was refreshed/restarted and new test conversations were initiated.
No visible mismatches, broken references, or configuration issues were identified.
Note: The agent flows were executing successfully during earlier testing sessions, and these runtime errors started appearing only within the last day, without any changes being made to the affected topics or conversation flow.
Impact
Due to the inconsistent occurrence of TopicNotFound and SystemError runtime failures, the agent conversation flow cannot be tested reliably in the Copilot Studio Test pane, which is currently blocking validation of ongoing development changes. Has anyone encountered similar behavior in Copilot Studio? Given that no obvious issues were identified in the topic configuration or flow logic, is it a studio platform glitch, are there any diagnostics or troubleshooting steps that could help determine the root cause? any guidance would be greatly appreciated.

Report
All responses (
Answers (