Lesson 108: Customer-visible failure observability
Focus
Bias toward observable metrics: latency, cost, escalation rate. Token Customer-visible failure observability:108 keeps neighbouring lessons differentiable.
Key ideas
- Thread: Customer-visible failure observability · drill v8 · spin
411229. - Habit: attach a trace_id to every completion you would paste into an ops dashboard.
- Guardrail: add one RACI bullet for prompt or index changes before tomorrow's standup.
Deep dive notebook
Synthetic drill artefacts
Agent choreography
1. Observe bucket `TENANT-3`
2. Max steps `7`
3. Tools: `retrieve, escalate_human, log_decision`
4. Stop: spend cap | keyword `STOP-10`
Practice
Practice Attach rollback steps if cost-per-request crosses your guardrail. — 108 Bump 30.