Availability and customer journeys
Review checks for important pages, application actions and dependencies. Define what a failed or slow result means for the people using the system.
Services
Build a clearer path from failure to recovery.
Start with the problem your team needs to detect or understand.
Logs, metrics and traces
Identify the information needed to investigate the chosen problem. Connect timestamps and affected components where practical, while limiting sensitive data and unnecessary collection.
Alert review and routing
Review thresholds, repeat notifications, recipients and missing coverage. Separate items needing prompt action from information that belongs in a later review.
Runbooks and recovery exercises
Document investigation steps, access owners, recovery decisions and checks after restoration. Use an agreed controlled exercise to find gaps without causing an unplanned outage.
Post-incident improvement
Reconstruct the timeline, identify contributing conditions and choose a manageable set of follow-up changes with owners.
AI feature monitoring
For an application using AI, review failed requests, timeouts, retries and fallback behavior alongside the customer task. A successful API response alone does not establish a useful answer.
Discuss a focused review. Ongoing coverage, response hours and operational responsibility require a separate agreement.