"Cannot reach backend" on the Zendesk Tickets tab: Fix: Start the backend with cd backend && python -m app.main. Confirm it's running at http://localhost:8000/docs.
No results after running a batch test: Fix: Check that OPENAI_API_KEY is set in your backend .env file and that the key has available credit. Visit http://localhost:8000/docs and try the health endpoint to confirm the backend is responsive.
All tickets show "Failed": Fix: This usually means the OpenAI API returned errors. Check the backend terminal for error messages. Common causes: invalid API key, expired key, rate limiting, or insufficient quota.
Mock import works but real import fails: Fix: Confirm your Zendesk OAuth connection is active (check the connection panel). If it shows "Not Connected," reconnect. If the token expired, disconnect and reconnect.
Scores seem inconsistent across runs: This is expected. AI evaluations have some natural variance because GPT-4 responses are non-deterministic. Run the same ticket set multiple times and look at averages rather than individual scores.
Progress bar gets stuck: Fix: Refresh the page and try again. If persistent, restart the backend — a ticket may have caused an unhandled error in the simulation pipeline.
Comments
0 comments
Please sign in to leave a comment.