Find the next step when something goes wrong

Start with the visible error and the task you were trying to complete. Check account, selected tool, and content before repeating a failed action.

Symptom	First checks
Sign-in repeats	Note where the redirect fails and check the intended account
A model or tool is locked	Read the access message and check the account’s plan
Reply has the wrong format	Check the reply selector; it applies to the next message only
A response is unhelpful	Add audience, goal, format, and relevant source material
A file is rejected	Read the type, size, or storage error; wait for processing
A chat seems missing	Check account, search filters, project, and Archived
Library seems empty	Clear filters and check the original conversation or Files
Writer work seems unsaved	Keep the editor open and check the Saved indicator
A visual omits information	Make relationships explicit and inspect every label
A scheduled task runs at the wrong time	Check the UTC time shown in its form
A connector is unavailable	Check setup requirements and permitted capabilities
An export is incomplete	Open the file and check selected content, format, and layout
An API call is rejected	Check key status, scope, model ID, schema, and response code

These are diagnostic starting points, not a guarantee that each error is solved by one action. Do not refresh away from unsaved work as a first step.

Describe an issue clearly

Record the tool, action, time, exact message, expected result, and actual result. Include a screenshot with private content hidden and a small synthetic example if it reproduces the problem.

Use the product’s current help or support destination. A support address, response-time commitment, and escalation process have not been confirmed for this documentation edition.