ExplainerAI in production
Muse Wonder's Active Memory needs an operational test
Test selective recall, corrected state, tool-failure recovery, clarification, and the independent destination result before trusting a long-running agent.
Articles
Guides, news and explanations about AI and reliable systems.
GuideObservability
Measuring streaming LLM latency from first text to completed work
A client-side SSE probe separates first answer text from normal completion. Fixed arrivals and outcome counts show whether a serving change delivers more useful answers.
GuideObservability
Measuring a latency SLO with Prometheus histograms
Build a latency SLI from Prometheus histogram buckets, include failed requests, and compare classic and native queries with a worked search-service example.
GuideObservability
Why a Prometheus error-rate query can return no data
Zero errors, no traffic and absent series produce different PromQL results. Synthetic tests show how ratios, fallback zeros and absence checks behave for a checkout service.
GuidePlatform engineering
Disruption budgets and the capacity left behind
Plan an inference-service node drain with a scoped disruption budget, replacement-capacity checks and a practical maintenance worksheet.
GuideSLOs & reliability
Release gates that hold up under incident pressure
An error-budget hold needs a route for an urgent repair. Fresh measurements, artifact-specific exceptions and recovery criteria determine how that route works.
CommentarySLOs & reliability
What an exhausted error budget changes
An error-budget policy connects measured reliability to release and planning choices, including a repair bundled with an unrelated feature.
GuideSLOs & reliability
Error-budget template: calculations and release-policy fields
Worked request and time-based budgets accompany an SLO record, release-policy template and exception log, keeping each calculation in its original units.
GuidePlatform engineering
What a canary deployment actually tests
Traffic assignment determines which behavior reaches a canary. Promotion also depends on the baseline, service objectives and state left behind by rollback.
ExplainerSLOs & reliability
Choosing an SLO for the work users need completed
A queued upload illustrates the choice of SLI, target and window, including why endpoint success can miss the user’s unfinished task.
ExplainerSLOs & reliability
How SRE feedback loops reveal work that moved out of view
Alert cleanup can reduce interruption or conceal a recurring failure. Customer reports and correction effort help distinguish the two results.
ExplainerSLOs & reliability
Error-budget calculations, balances and burn rates
Worked request and time examples explain error-budget units, remaining allowance and burn rate, plus the release policy that gives them a practical role.