Intelligent summaries
CircleCI’s intelligent summaries provide contextual help with build failures. Select Explain this error from a failed job’s step output to see an AI-generated summary of the error, along with suggestions on how to fix it. Use intelligent summaries to speed up the process of resolving common issues without needing to dig into documentation, or other resources.
Enable intelligent summaries
The intelligent summaries feature is disabled by default. You can opt-in to the feature, as follows:
| Only organization administrators can enable or disable intelligent summaries. |
-
From the CircleCI web app select your organization card.
-
Select Organization Settings from the sidebar.
-
Select Advanced from the menu.
-
Use the "Enable intelligent summaries of build failure messages" toggle to enable or disable the feature.
Use intelligent summaries
The Explain this error button is available at the step output level in the CircleCI web app. You can locate the option by clicking through to the failed step from your homepage, as follows:
-
Select your organization card from the homepage.
-
Select Pipelines from the sidebar.
-
Locate your pipeline that has a failure, and use the arrow in the Status column to expand the failed workflow.
-
Select the name of the failed job.
-
From the job view, with the Steps tab selected, you are viewing the step output. Locate a failed step and select Explain this error.
-
A panel appears on the right side of the screen with the intelligent summary, which includes a description of the error and steps to fix it.
If you see a banner stating that "Intelligent Build Failures are Disabled", an organization administrator can enable the feature as outlined in the previous section.
Privacy and security
Intelligent summaries works by sending an error message along with build failure logs to a third-party LLM. Prior to submitting the error message and logs to the LLM, CircleCI passes the message through a filter to remove any sensitive information, such as, access keys and passwords.
OpenAI is used as the LLM for this feature. If this changes, this page will be updated to reflect the change, and there will be an announcement on our community forum.
The data we send to the LLM is not used for training purposes and no data is saved by the LLM. No personal information is sent to the LLM.