The Series

Above the Model

The engineering that matters happens one level above the model.

Models get more capable every release, and none of that decides whether your product improves next month. That part is engineered one level up: turning vague demand into briefs agents can execute, trusting a loop's output before it changes anything, building the instrument that feeds the system evidence about itself, and containing behavior that raw capability does not fix. Code became a byproduct. The scarce input is well-formed intent, and the system around the model is the part you actually own.

Four essays, written from a year of daily practice, in reading order.

A minimal Zen-style illustration of tangled demand passing through three torii gates and emerging as ordered parallel lines feeding clean briefs.

1. Intent Engineering

Working Above the IDE

April 25, 2026

Code became a byproduct. The scarce input is well-formed intent. Notes on a year of building and daily-driving a post-IDE workstation.

A minimal Zen-style illustration contrasting a single sketchy checkmark with a ring of many loops, one incomplete.

2. Loops Need Proof, Not Vibes

Why "It Passed" Isn't Evidence

May 30, 2026

The industry found its word for how agents improve, the loop. What the conversation skips is the step that makes a loop trustworthy when the exit criterion is semantic and the agent underneath is nondeterministic.

A minimal Zen-style illustration of a large enso circle holding a small inner circle, with thin conduits carrying its output to documents outside.

3. The Model Is Not the System

Build the Structure, Don't Wait for the Model

June 27, 2026

One camp waits for the next model to make agent problems disappear. The other engineers the structure around the model, including the instrument that feeds the system data about itself, so it can evolve on evidence instead of faith.

A minimal Zen-style illustration of a bold line climbing steadily while a ribbon undulates below, running straighter where it passes through an open containment frame.

4. Capability Compounds, Discipline Doesn't

Why Smarter Models Still Need Guardrails

July 25, 2026

Every release makes models more capable. It does not reliably make them better behaved. Those are different axes, and the industry's roadmap quietly assumes they are one.

The Field Notes Behind the Series

A minimal Zen-style illustration of one operator coordinating thirteen parallel flowing streams, one broken, converging into a single check.

Running Thirteen AI Agents in Parallel

An Operations Post-Mortem

March 28, 2026

One coordinating session drove thirteen parallel agent work units through analysis, implementation, measurement and benchmarking. Here is everything that broke, and the operating rules that came out of it.