← All writing

Strategy

Leadership thinking, philosophy, and governance approaches

Give the agent superhuman work

A French cloud CEO describes moving seventy developers into agent-assisted work and converges on conclusions close to this series. The most valuable work had never been worth a person's time.

When writing code is no longer the developer's main job

When agents handle implementation, developers move towards problem framing, verification, operations, and responsibility. The apprenticeship model must adapt too.

AI-assisted development is still software development

AI-generated code does not reduce engineering responsibility. Higher throughput demands stronger review, security, identity, and deployment controls.

The workflow is now built around AI

Why frontier development has moved beyond assistants and prompts towards harnesses, persistent goals, and systems built around agents.

Stop automating broken processes

Why automation amplifies problems instead of solving them. A broken process should be redesigned, not automated.

Governance without bureaucracy

Designing governance processes that create alignment without slowing teams down. Good governance enables speed, not friction.

When security fights users, users win

How user experience shapes security outcomes more than most controls do. UX is part of the security model, not an afterthought.

Simplicity is the strongest security control

Why reducing complexity is often more effective than adding security layers. Complexity breaks people and systems alike.