From AI experiments to organizational shift: Lessons from Mercari’s transformation (Michael Galloway and Snehal Shinde)

Mercari’s transition to an AI-native organization reveals that AI adoption is not a finish line but a catalyst for exposing deep-seated organizational bottlenecks. While AI significantly accelerates coding, it often hits friction in legacy systems, manual approval gates, and fragile CI/CD pipelines, leading to a "W curve" of productivity dips. Successfully scaling AI requires shifting focus from local code generation to system-wide visibility, addressing the "awareness gap" through community-driven workshops, and implementing agent-spec driven development to integrate AI at the start of the product lifecycle. Ultimately, sustainable progress depends on balancing total faith in the AI-native destination with brutal honesty about current operational realities, ensuring that infrastructure, guardrails, and workflows evolve to support automated execution rather than just augmenting existing, broken processes.