We use Trigger.dev for AI streaming tasks and would love built-in support for resumable streams.
Currently if a client disconnects mid-stream (tab backgrounded, network flap, page refresh), you either lose in-flight data or have to build custom replay logic. This is especially painful for long-running AI tasks where you're streaming tokens or progress events.
The new Durable Streams protocol solves this at the transport layer - HTTP-based streams with offset tracking that let clients resume from exactly where they left off. It's built by the Electric team and designed specifically for this use case.
Would be great to see Trigger.dev adopt this for:
Realtime log streaming
AI token streaming with automatic retry/resume
Job progress events
Any long-running task output
The protocol is simple (standard HTTP + offset headers) and CDN-friendly, so it should integrate well with existing infrastructure.
In Review
π‘ Feature Request
23 days ago

jbingen
Get notified by email when there are changes.
In Review
π‘ Feature Request
23 days ago

jbingen
Get notified by email when there are changes.