Overview
Most issues come down to access, credentials, or provider settings. This page helps you diagnose problems step‑by‑step and points you to the right place to fix them.
- Access: Make sure users have the correct roles.
- Connectivity: Verify provider name, model, and API key.
- Observability: Use Usage Logs for quick clues and filters.
Real‑life example: Subjects not generating? Check that the model name is supported and the token limit isn’t too low.
How It Works
The app calls your chosen AI provider using your settings and merges context. If a call fails, it may return a simple fallback suggestion.
Step-by-Step Guide
- Check user roles: “AI Mail User” for access, “AI Mail Manager” for setup.
- Verify Settings → AI Suggestions: provider, API key, model name.
- Test with a simple prompt on a common record.
- Open Usage Logs for details. If needed, enable debug logging in staging only.
Fields Table (Helpful checks)
| Field Name | Description | Example | Use Purpose |
|---|---|---|---|
Roles |
User must be allowed to use AI |
AI Mail User |
Unlock AI buttons in the UI |
Provider & Key |
Correct credentials |
openai + secret |
Confirm connection succeeds |
Model name |
Supported by your provider |
gpt-4o-mini |
Ensure the selected model exists |
Field name explanation
Roles
Give users the right access so AI buttons appear.
Provider & Key
Confirm the service name and secret are correct.
Model name
Use a supported model at your chosen provider.
Common Mistakes
- Using a deprecated model name.
- Forgetting to save settings after edits.