We need to move away from executing long-running tasks (like story generation) within the main request handler to avoid timeouts on Vercel (10s/60s limits).
Requirements:
- Evaluate Inngest vs QStash (or others).
- Implement a background job for call LLM service.
- Implement polling or websocket/SSE to update UI when story is ready.
We need to move away from executing long-running tasks (like story generation) within the main request handler to avoid timeouts on Vercel (10s/60s limits).
Requirements: