This message appears when the backend's call to Anthropic returns an error, but the error message itself isn't user-friendly enough to show as-is. Here's how to narrow down what's actually going on.
1. Check your token balance
The most common cause: you ran out of AI tokens. Open your account page — the meter at the top shows your remaining balance. If it's zero, that's your answer; top up or upgrade.
2. Check the backend health
Visit api.screensolo.ai/api/v1/health. The JSON response should say:
{ "status": "healthy", "anthropic": "reachable", "stripe": "reachable", ... }If anthropic says anything other than reachable, the issue is on our side. Wait a few minutes, refresh, retry the AI action.
3. Check the specific error in the toast
Builds from 13 May 2026 onwards forward the *actual* error message from Anthropic into the toast — so instead of "AI service encountered an error" you should now see the real reason. If you're still seeing the generic message, update to the latest build.
4. Common specific Anthropic errors
- "Image specified using image/png but appears to be image/jpeg" — was a bug, fixed in backend deploy May 13. Should not appear in current production.
- "Rate limited" — too many requests in a short window. Wait 60 seconds and retry.
- "Invalid request: content blocked" — Anthropic's safety filter rejected the request. Usually means the keyframe contained something the model declined to analyse (NSFW, harmful content). Try a different keyframe.
5. None of the above?
Email support with:
- A screenshot of the toast
- The action you took (which AI feature, which session)
- Approximate timestamp
We can pull the corresponding row from our audit log and tell you exactly what went wrong.