How AI has changed my engineering leadership role

What I took from it

• Stay hands-on, selectively: small, high-leverage fixes that make the team safer, not feature work that competes with them.
• Example: after a bad release, Roche built a simple feature-flag system with an AI agent in a few hours, and the team adopted it.
• Faster onboarding: AI explains unfamiliar code, so people ramp up quicker.
• Less resistance to new tech: ramp-up costs are lower, so new tools are easier to justify.
• Lower bus factor, not zero: AI shows what code does, not why it was built that way. Keep writing decisions down.