Provider Strategy
Provider-specific code should remain isolated behind stable product contracts so the application can adapt to provider changes without rewriting user-facing flows.
Overview
Section titled “Overview”The provider layer is a production risk area because model APIs, latency, disclosure requirements, and output behavior can shift.
Decision
Section titled “Decision”- Keep provider details behind a narrow service boundary.
- Normalize provider outputs before they reach product routes.
- Track latency and timeout behavior separately from product success.
- Make queueing a first-class architecture concern before payment hardening.
Dependencies
Section titled “Dependencies”This decision connects directly to the Vercel timeout and queue-required risks.