A row of identical toy aeroplanes rolling off a fast conveyor belt onto an unmarked snowy mountainside, where a roped climber seen from behind plants marker flags ahead of them, the whole scene drawn with wide empty margins on every side.

martinfowler.com (Rachel's Ramblings)

Why are good developers so wary of agentic development?

Rachel Laycock · 29 October 2026

Company story · agents · roles · Changing roles

Takeaway

Thoughtworks' chief technology officer Rachel Laycock says experienced developers distrust AI agents because built software is not yet working software, and they should help design the new practices.

Summary

Rachel Laycock, chief technology officer of the software consultancy Thoughtworks, asks why many of its best developers hesitate to let AI agents do all the coding. She draws on a week-long trial in which a Thoughtworks team built an airline application without writing any code by hand. She concludes that the hesitation is a reasonable concern about production, not a fear of change.

Key points

Giles Alexander, Thoughtworks' regional chief technology officer for Europe, had senior engineers, infrastructure consultants and product people build an airline application using only AI agents. The trial ran twice with different teams.
The teams estimated they spent about two thirds of their time at the whiteboard and one third on the tests, contracts, architecture checks and delivery pipeline the agents worked against.
The application ran, but the build was frequently broken, and the product person committed more feature code than anyone else.
Laycock asks leaders to give cautious developers time and permission to experiment, and not to judge success by how much code the agents produced.

Implication

An engineering team adopting AI agents can treat its most cautious senior developers as the people who design the tests and delivery checks for agent-written code. That is an inference from the trial; Laycock's own claim is that their caution is about production, not about change.

Plays

Engineering

Before AI agents write a feature's code, have senior engineers write the tests, interface contracts and architecture checks that the agents' code must pass in the delivery pipeline.

Engineering

Run a week-long trial in which AI agents write all the code, and judge it by what the team learns, not by how much code the agents produced.

Derived by Working Surface from the article; more in the Playbook. Source line: Good developers’ caution about agentic development comes from knowing that built isn’t the same as working in production.

Source issue

30 October 2026: as AI agents do more of the building, teams decide on purpose which work people keep and set up tests, usage figures and baselines before handing over more.