[Feature] Automatic retry for temporary AI request failures

Clear and concise description of the problem

First of all, I really appreciate the work you guys are doing on AIRI. The project has been getting some really interesting improvements, and I’ve genuinely enjoyed trying it out. I also really appreciate that you’re considering the Fish Audio API key option I suggested earlier — it’s really cool to see the idea being taken into consideration. ❤️

I had another feature suggestion that might make the experience a bit smoother:

Automatic retry for temporary AI request failures

Sometimes an AI request can fail because of temporary server demand, rate limits, network issues, or a model/service being temporarily unavailable. In those situations, the user has to manually send the same message again.

Could AIRI automatically retry the request when the error appears to be temporary?

Something like:

"Request fails → wait a few seconds → retry → retry a limited number of times"

It could ideally respect the provider's "Retry-After" value when available, or use exponential backoff such as 3s → 6s → 12s.

Permanent errors like invalid API keys or invalid requests shouldn't be retried automatically, of course.

I think this would be a small feature that could make AIRI feel more reliable during temporary service spikes, while saving users from having to resend longer messages manually.

Thanks again for all the work you put into AIRI! Really looking forward to seeing where the project goes. ❤️

Suggested solution / Ideas

No response

Alternative

No response

Additional context

No response

Validations

• Follow our Code of Conduct
• Read the Contributing Guide.
• Check that there isn't already an issue that request the same feature to avoid creating a duplicate.