Ideas and field notes

Practical writing on building AI that ships, from the team doing the building.

The state of applied AI in 2026

Where applied AI actually stands this year, past the hype cycle and into what teams are really shipping.

Plans, tools, guardrails, and observability: the unglamorous work that makes agents dependable.

A practical retrieval pattern that stays accurate, cited, and honest about what it does not know.

Why your eval harness matters more than your model choice, and how to build one that holds.

Why the systems you depend on should run in your cloud and belong to your team.

How to turn a single AI win into a platform that compounds across the business.

Practical writing on building AI that ships. Monthly, no fluff.