Jekyll2026-10-09T18:28:59+00:00https://productionnotes.dev/feed.xmlProduction NotesProduction engineering notes on enterprise AI agents, workflows, and SRE — by Sabith K S, Principal Engineer at StackGen.Sabith K SAI Agent Evals in CI/CD: Flakes, Retries, and Missing Results2026-10-07T23:30:00+00:002026-10-07T23:30:00+00:00https://productionnotes.dev/blog/ai-agent-evals-cicd-flakesSabith K SMulti-Agent Handoff Testing: Catch Context Loss Before Production2026-10-07T19:00:00+00:002026-10-07T19:00:00+00:00https://productionnotes.dev/blog/multi-agent-handoff-testing-context-lossSabith K SHow to Test AI Agent Loops Without Overfitting2026-10-07T15:30:00+00:002026-10-07T15:30:00+00:00https://productionnotes.dev/blog/test-ai-agent-loops-with-evidenceSabith K SDeterministic Checks vs LLM-as-a-Judge for Agent Evals2026-10-07T02:00:00+00:002026-10-07T02:00:00+00:00https://productionnotes.dev/blog/deterministic-checks-vs-llm-judgeSabith K SThe AI Agent Testing Pyramid: What Belongs in CI?2026-10-06T22:00:00+00:002026-10-06T22:00:00+00:00https://productionnotes.dev/blog/ai-agent-testing-pyramidSabith K SAI Agent Evaluation: Why the Grader Must Be Separate2026-10-06T16:30:00+00:002026-10-06T16:30:00+00:00https://productionnotes.dev/blog/ai-agent-evaluation-separate-graderSabith K SSame Incident, Five Models: Rank Investigators Without Mixing the Exam2026-09-18T22:00:00+00:002026-09-18T22:00:00+00:00https://productionnotes.dev/blog/same-problem-sre-model-bake-offSabith K SSession Efficiency Should Not Beat Accuracy2026-09-16T19:00:00+00:002026-09-16T19:00:00+00:00https://productionnotes.dev/blog/session-efficiency-should-not-beat-accuracySabith K SChat Completions vs Responses API (and Why Agents Feel Slow)2026-09-10T05:00:00+00:002026-09-10T05:00:00+00:00https://productionnotes.dev/blog/chat-completions-vs-responses-apiSabith K SHybrid Plan Mode: Smart Planner, Generate Digs2026-09-10T04:00:00+00:002026-09-10T04:00:00+00:00https://productionnotes.dev/blog/hybrid-plan-smart-planner-generate-digsSabith K SAI Agent Runtime: What to Measure Before You Buy2026-09-10T03:00:00+00:002026-09-10T03:00:00+00:00https://productionnotes.dev/blog/ai-agent-runtime-what-to-measure-before-you-buySabith K SHow to Evaluate an AI Agent for Root Cause Analysis2026-09-10T01:00:00+00:002026-09-10T01:00:00+00:00https://productionnotes.dev/blog/how-to-evaluate-ai-agent-root-cause-analysisSabith K SReasoning vs Generate Models for Tool-Heavy Agents2026-09-09T23:00:00+00:002026-09-09T23:00:00+00:00https://productionnotes.dev/blog/reasoning-vs-generate-tool-heavy-agentsSabith K SBest Way to Debug a Multi-Step AI Agent in Production2026-09-09T21:00:00+00:002026-09-09T21:00:00+00:00https://productionnotes.dev/blog/best-way-to-debug-multi-step-ai-agentSabith K SRelative Efficiency Scores Lie When Absolute Wall Doubles2026-09-09T19:00:00+00:002026-09-09T19:00:00+00:00https://productionnotes.dev/blog/relative-efficiency-scores-lieSabith K SStop Retrying the Same Failed Observability Query2026-09-09T17:00:00+00:002026-09-09T17:00:00+00:00https://productionnotes.dev/blog/stop-retrying-the-same-failed-querySabith K SHierarchical vs Single-Agent on a Dual-Part Observability Job2026-09-09T01:00:00+00:002026-09-09T01:00:00+00:00https://productionnotes.dev/blog/plan-mode-merits-demerits-observabilitySabith K SObservability Tools Agents Actually Call on Triage2026-09-08T23:00:00+00:002026-09-08T23:00:00+00:00https://productionnotes.dev/blog/observability-tools-agents-actually-callSabith K SWhat Reasoning Models Write When You Ask Them to Triage2026-09-08T21:00:00+00:002026-09-08T21:00:00+00:00https://productionnotes.dev/blog/what-reasoning-models-write-on-triageSabith K SSix Model×Orchestration Combos on One Alert+Logs Job2026-09-08T19:00:00+00:002026-09-08T19:00:00+00:00https://productionnotes.dev/blog/six-model-mode-combos-alert-logs-benchSabith K S