A line chart of three costs over time. The 'cost of hardware' curve starts highest on the left and descends across all three eras. The 'cost of writing code' curve starts moderate, rises through the 1960s to peak at the 1968 software crisis marker (where it overtakes the hardware curve), plateaus through the 1990s, then drops at the 2020s marker as agents take over writing. The 'cost of checking meaning' curve emerges sharply at the 2020s marker, rising to become the highest cost on the right. Three vertical dashed markers label '1968 software crisis', '1990s complexity & scale', '2020s code agent crisis'. An annotation 'where the work now lives' points to the widening gap between checking-meaning (high) and writing-code (low) on the right side.

The Code Agent Crisis

Software engineering has had two crises before. They had the same shape: something became cheap, and the work moved to whatever was still hard. We are in the middle of a third.

May 10, 2026 · 7 min · Yann Régis-Gianas
A hand-drawn diagram read left to right. Dotted boxes hold artifacts: context, environment and goal on the left, new environment and verdict on the right. Solid boxes hold whoever acts on them: the engineer, and the deterministic checks and auditors. One heavy box holds the agent, Ctx times Env times Goal to Env prime. Thin arrows carry a verb, thick arrows carry an artifact, and nothing points back. A dashed boundary around everything reads AROUND THE AGENT: EVERYTHING YOU CAN ENGINEER.

Why a loop at all

An agent is a stochastic function with no memory. Everything you can improve is around it, which is why the durable engineering object is the loop, not the output. The argument from first principles, and the slogan that follows: fix the loop, not the output.

May 22, 2026 · 13 min · Yann Régis-Gianas
A hand-drawn diagram of the loop: five activity boxes in a ring, and at the centre a dashed box, the engineer's mental model, connected to four of them by two-way dashed arrows.

The loop I run

The concrete loop: five activities in a ring, three components the team owns, a persistent model of the engineer at the centre, and the routing rule that decides which layer a failure belongs to.

August 5, 2026 · 12 min · Yann Régis-Gianas

Keep the engineer in the loop

Autonomous cycles erode the understanding of the person accountable for the system. This essay makes that person a first-class component of the loop: a persistent model of what the engineer knows, consulted before the loop explains or asks anything, and updated every cycle. It is an objective the process works to keep, not a property it can prove it holds.

August 5, 2026 · 15 min · Yann Régis-Gianas

One lap of the loop, on a real project

In one recorded day, a proof assistant refused a specification its engineer had approved, and later certified two true theorems about a tool whose hidden words a forty-word dictionary read off in milliseconds. A different station of the loop caught each defect. This essay walks that day, tool by tool, screen by screen, against a public record.

August 5, 2026 · 29 min · Yann Régis-Gianas

Hello, World

Introducing this blog — why I’m writing about software engineering in the agent era.

April 2, 2026 · 1 min · Yann Régis-Gianas