Your junior engineers are getting faster but are they getting better?

Engineering leader Garima Agarwal found that junior engineers using AI coding tools shipped faster without understanding the systems, and rebuilt onboarding to restore that understanding.

Garima Agarwal leads an engineering team and writes in LeadDev, a publication for engineering managers. After six months of mandatory AI coding tools, junior engineers on the team had shipped 14 features against a service and could not explain how it worked. Agarwal changed onboarding over the next six months without restricting the tools.

• A junior engineer was called to a production incident in a service he had written a feature for two months earlier, and did not know where to start.
• Onboarding tickets are split into output tasks, where AI tools are encouraged, and learning tasks, which a junior first attempts without AI and then explains in writing.
• Once a week, each junior engineer spends 15 minutes explaining the decisions in one merged pull request to a senior engineer.
• New engineers spend their first two weeks reading the codebase without AI tools. By Agarwal's count, the time before a junior could handle an incident alone fell from seven months to four.

An engineering manager can ask a junior engineer to explain a service without a laptop, to learn whether fast delivery hides a gap in understanding. This test is the author's own recommendation; the results come from one team and nobody outside it has checked them.

Derived by Working Surface from the article; more in the Playbook. Source line: The reasoning never passed through him.

9 October 2026: when AI agents take on routine work, teams write down in advance what a person still decides, put those rules where the agent reads them, and keep the people who approve able to understand the work.