Are technical design reviews disappearing?

Diego Quiroga, an engineering leader at Microsoft, argues that AI coding agents speed up coding but do not remove the need for a short design review on risky changes.

Diego Quiroga, a senior engineering leader at Microsoft, writes in LeadDev about a team whose feature went straight from a product requirements document to code written by AI coding agents. The pull requests delivered the requested behaviour, but reviewers found assumptions taken from single codebases and missing precautions on critical paths. A two-page document on intent and approach let the team revisit those assumptions before the change went further.

• The feature touched several services and data models, and the usual technical design review through a request for comments had been skipped.
• Reviewers asked the author for a two-page document, which slowed the feature but brought it closer to the organisation's architectural principles.
• Quiroga keeps design reviews for changes that cross service or team boundaries, change stored data or public contracts, add security or reliability risk, or are expensive to reverse.
• His design brief covers the problem and constraints, the proposed approach, the alternatives, the trade-offs, failure modes and reversibility, and open assumptions.

Quiroga suggests that engineering teams keep a two-page design review for risky changes made with AI coding agents, and relax it once agents prove sound on a class of change. Using an agent is not, in his view, a reason on its own to require a review.

Derived by Working Surface from the article; more in the Playbook. Source line: Code that delivers the requested behavior is not necessarily the right technical solution to the problem.

28 October 2026: as AI agents make code look finished sooner, teams are moving the human check to before the build and keeping a person accountable for the changes that are hardest to undo.