Martin Fowler's May 27 Fragments brings together four arguments with direct implications for teams working with AI agents. All four are worth covering.

Ian Johnson: build quality gates before releasing the agent

Ian Johnson published a series about restructuring a gnarly codebase: three months, 258 commits, moving from a Laravel monolith with no tests to an application with automated quality gates and an AI agent shipping production code with minimal supervision.

The insight Fowler highlights is about the transition from in-the-loop to on-the-loop:

"For the first two months of this project, I used Claude Code with auto-approve turned off. Every file edit, every terminal command, every change… I reviewed it before it executed. The results were good. The code was clean. But I was doing most of the thinking and half the typing. The agent was a fancy autocomplete with better suggestions."

Ian Johnson

Manual review of every change is not how you build trust in the agent. Trust comes from building the structure that ensures the agent will do the right thing, then stepping back. The sequence: characterization tests first, static analysis, architectural patterns that make things flow correctly. Fowler notes this is exactly the sequence he would use himself.

Adam Tornhill: roughly 2 hours of cognitive endurance

Adam Tornhill observes that agentic work has a decision density that is mentally more expensive than it appears. The estimate is roughly two hours as a sustainable limit, not a full day of work.

The implication: adding more parallel agents does not solve the problem, because the bottleneck is the coordinating engineer's cognitive capacity, not available processing volume. The solutions are smaller tasks, automation, and verification mechanisms, not more parallelism.

NHS: closing open source repositories

NHS (UK National Health Service) closed open source repositories citing LLM threats to code security. The UK Government Data Services countered directly: making code private reduces scrutiny and coordination without eliminating the underlying vulnerabilities. Private code obscures the problem, it does not solve it.

Economic impact: the data that will get worse before it gets better

Labor market data shows graduates with high AI exposure experience a 6.6% employment drop versus 1.5% among less-exposed peers. The historical argument that major technological advances rarely cause long-term unemployment exists, but current data points in a different direction. Fowler's hypothesis is that the full consequences will likely only be visible during a recession.

For leadership teams, the four points have distinct practical implications. Johnson's sequence is an adoption framework. Tornhill's endurance limit is a sprint planning criterion. The NHS posture is a security case to study. The economic data is context that will remain relevant across the next few planning cycles.

Fonte: Fragments: May 27