Writing
Field notes from production agent infrastructure
Long-form, code-first writing on cutting LLM costs, building evals, and making codebases ready for agents. No thought-leadership fluff.
- 6 min readagentic-devinfrastructure
The shippable agent-month: a working definition
The mythical agent-month is a critique of naive AI productivity claims. The shippable agent-month is what one well-instrumented agent actually delivers.
Read - 3 min readdevexplatform
Standardizing AI coding tools without killing autonomy
How DevEx leaders turn a Claude Code, Cursor and Copilot free-for-all into a golden path: shared commands, MCP access and review hooks, with no single mandate.
Read - 3 min readctostrategy
The “add AI” mandate: a CTO’s first 90 days
The board wants AI in the product before the bets are proven. A sequencing playbook for CTOs: what to ship first, build vs buy, and keeping cost forecastable.
Read - 3 min readsecurityai-code
Securing AI-generated code: a checklist for security leaders
AI writes code faster than anyone can review it. A checklist for CISOs and AppSec leads covering AI-generated code, prompt data leakage and injection surface.
Read - 3 min readevalsquality
You shipped AI features with no evals. Here’s how to fix that.
Most teams shipped production AI with no evaluation. How to add evals incrementally, without stopping the roadmap, so model changes stop being blind deploys.
Read - 3 min readagentic-devcodebase
Is your codebase agent-ready? A scoring rubric
Agents thrive in some codebases and thrash in others. The rubric we use to score how ready a repo is for agentic development, and what to fix first.
Read - 4 min readllm-costsobservability
How to cut LLM costs without losing quality
A vendor-agnostic playbook for finding and removing wasted production LLM spend: model routing, caching, batching, and observability to keep it gone.
Read