<Notes of dev/>

AI Agents

4 posts

Three nested concentric frames — a single dot (context) inside a loop arrow (loop) inside a network of nodes (graph) — showing each layer wrapping the one before it.

Context, Loop, and Graph engineering

August 4, 202614 min read

They are nested layers, not rival techniques. What each one does, when it's enough, and when to reach for the next.

A marble observatory on a plateau at golden dawn with a giant sundial and armillary sphere.

The AI agent ROI playbook

July 21, 202628 min read

The measurement machinery for AI agents: a self-test against your own past, a PM dashboard you refresh in one command, and the role-by-role setup that makes ROI real.

A colossal marble balance scale in a bright valley: a pan heaped with gold floats weightless while one glowing stone tips the beam.

How to measure AI agent ROI

July 4, 202625 min read

Measuring AI agent ROI without fooling yourself: outcomes over activity, how to catch an agent going bad, and why product teams and outsourcing teams count it differently

A grand marble temple on solid bedrock in a sunlit valley, with small scattered ruins fading into morning haze.

The complete AI coding agent setup for Claude Code and Copilot

July 24, 202623 min read

The files, context, and workflow that turn an AI coding agent into consistent, reviewable output — set up once, for both Claude Code and Copilot.