How-To

Practical methods you can take back to your team.

A zero fallback can hide missing telemetry. Use a small PromQL test matrix to check what your error-rate alert can and cannot establish.

September 25, 20267 min read

Plan an inference-service node drain with a scoped disruption budget, replacement-capacity checks and a practical maintenance worksheet.

September 24, 20267 min read

Build a prompt-caching pilot that counts retries, failed work and unknown charges, with a local calculator and sample ledger.

September 23, 20267 min read

Build repeatable cases and check recovery independently, with a runnable Python grader.

September 22, 20269 min read

Use instrumentation and context propagation to connect the evidence in an investigation.

March 24, 20264 min read

Clarify ownership across shared platforms, service reliability, and incident response.

March 24, 20264 min read

Separate a completed task from evidence that the underlying risk was reduced.

March 24, 20265 min read

Follow evidence through technical failure and incident response, then test whether the proposed correction changes the mechanism.

February 13, 20264 min read

Use a curated notebook to compare runbooks and incident evidence, while keeping citations, freshness, and operational authority in view.

February 12, 20264 min read

A design guide for an incident assistant that handles duplicate events, partial failures, and reviewed AI summaries across tools.

April 6, 20254 min read

Build a view that answers who is affected, what changed, and where to investigate, then test its queries and missing-data behavior.

April 3, 20253 min read

A reusable prompt pattern for incident analysis, with checks for missing facts, unsafe recommendations, and unsupported certainty.

March 31, 20253 min read