Observability Design for the AI Era — Application / Infrastructure / CI / LLM, Each in Its Own Shape (Part 1)
Scored daily by a customisable AI persona to surface the most relevant engineering leadership news.
Observability design for AI era is highly technical, novel, and directly matches reader's interests.
airCloset CTO Ryan Tsuji splits observability into four shaped surfaces—application (OTel + Loki + Tempo), infrastructure (metrics), CI (logs + alerts), and LLM (metrics + structured records)—to make telemetry AI-consumable, avoiding the context-window drowning and hallucination that raw logs cause. Each surface is optimized for specific AI queries: real-time production exploration, resource health, breakage history, and cost/usage tracking. The key discipline is uniform log/trace shapes across all services, enabling AI tools like MCP to cross-service query with patterns like `{service_name="<service>"} |~ "error"`.