Writing / Guides / 5
The long versions.
Each guide pulls a cluster of essays into one piece you can read end to end, then hands you the essays for the detail.
AI Agent Reliability in Production: What Breaks and What Fixes It
Every way an AI agent fails in production, from bad tool calls to blown error budgets, with the pattern that fixes each one and the essay that goes deep.AI Agent Security: A Guide to Containing What Agents Can Do
Thirteen essays on AI agent security in reading order: trust boundaries, per-call authorization, scoped identity, egress checks, sandboxes, supply chain gates, and audit records.Context engineering for AI agents: the production guide
One reference for the whole discipline: what goes into an agent's context window, what gets pushed out, and the patterns that keep long runs cheap, true, and auditable.LLM Cost Optimization for Agents: A Production Guide
Every lever that cuts an agent's LLM bill, in the order they pay off: routing, effort, prefix and result caching, context control, per-tenant caps and tracing.LLM Evals in Production: From a CI Gate to a Living Eval Set
How to build LLM evals that hold up in production: a CI gate, a calibrated judge, agent tests that see the path and the conversation, a set that feeds itself from traffic.