85. An external AI provider becomes unreliable. What should your team's continuity plan address?
HardAI engineering leadership
Managerial
Show model answer
Identify affected tasks, detection signals, ownership and which activities must pause or degrade. Any fallback must meet the relevant behavior, data, location and contractual requirements rather than merely returning a response.
Rehearse feasible alternatives, preserve user work and coordinate truthful status communication. Verify restoration before returning to normal operation, and avoid assuming that providers are interchangeable or that an untested fallback will behave safely during an incident.
What interviewers listen for
- Define task-specific degrade or stop behavior
- Validate fallback requirements
- Rehearse recovery and truthful communication