How Maven Clinic rebuilt engineering for an AI-native world: 5 leadership lessons

Dan Feng from Maven Clinic describes how his company transformed into an AI-native organization over two years. The talk focuses on three key areas: internal AI tool adoption, external product integration, and cultural/process changes. Feng argues that forcing AI usage across all employees is critical, with clear expectations set for those slow

to adopt. The most striking insight is how AI has fundamentally changed engineering roles. Senior engineers now prefer using AI themselves rather than delegating implementation work to juniors, reducing overhead. This shifts hiring criteria toward product-savvy engineers who can handle ambiguity, rather than pure coding skills. Performance reviews now explicitly ask what

each employee has done with AI. Process changes are equally dramatic. Feng advocates abandoning 3-6 month planning cycles in favor of two-to-four-week sprints, arguing that AI model capabilities evolve too quickly for medium-term planning. He also describes a radical shift in code review practices: allowing self-approval for simple PRs, limiting PRs to

500 lines, and using stacked PRs to avoid bottlenecks. The talk also covers reliability strategies for AI systems, including multi-model consensus for high-risk decisions, extensive integration testing with multiple runs per test case, and post-launch conversation monitoring with dedicated human reviewers. Feng acknowledges they haven't solved end-to-end automation or AI-driven monitoring yet.