AI Features
AI Error Explain
AI-powered analysis of failed webhook deliveries using GPT-4o-mini. Get root cause analysis, suggested fixes, and severity ratings.
Mitte uses GPT-4o-mini to analyze failed webhook deliveries and provide actionable debugging insights. Instead of manually interpreting HTTP status codes and error messages, let AI explain what went wrong and how to fix it.
How It Works
When a webhook delivery fails, you can click Explain with AI in the log inspector. Mitte sends the following context to GPT-4o-mini:
- HTTP status code and error message
- Response body from your target server
- Request duration and attempt number
- Detected webhook source provider
The AI returns a structured analysis:
| Field | Description |
|---|---|
| Summary | A one-line description of the problem. |
| Explanation | Detailed analysis of what went wrong and why. |
| Suggested Fix | Concrete steps to resolve the issue. |
| Severity | low, medium, high, or critical. |
Caching
AI explanations are cached in Redis for 1 hour. Subsequent views of the same failed delivery will load the cached result instantly without consuming an AI quota.
You can force a fresh analysis by clicking Regenerate — this bypasses the cache and uses a new AI quota.
Daily Limits
| Plan | AI Explains / Day |
|---|---|
| Free | 5 |
| Pro | 100 |
The quota resets daily at midnight UTC.
Example
For a failed delivery returning 502 Bad Gateway:
Summary: Target server returned 502 Bad GatewayExplanation: Your target server at
https://api.myapp.com/webhooks is returning a 502 status, which typically means a reverse proxy (like Nginx or Cloudflare) couldn't reach the upstream application server. This often happens when the app process has crashed or is restarting.Suggested Fix: Check that your application server is running and healthy. Review your reverse proxy logs (Nginx, Cloudflare, etc.) for upstream connection errors. Consider adding a health check endpoint.Severity: high