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

Engineering Enablement by DX

From AI experiments to organizational shift: Lessons from Mercari’s transformation

0:00

-44:38

From AI experiments to organizational shift: Lessons from Mercari’s transformation

What Mercari learned after mandating 100% AI adoption—and why faster code generation didn’t automatically lead to faster software delivery.

Justin Reock

Jun 15, 2026

Listen and watch now on YouTube, Apple, and Spotify.

Michael Galloway leads Platform Engineering at Mercari, while Snehal Shinde leads Cost and Performance Engineering. Together, they have been at the center of Mercari’s effort to become an AI-native company.

In this session from DX Annual, Michael and Snehal share what happened after Mercari’s CEO mandated 100% AI adoption across the organization. While AI accelerated code generation and increased engineering output, the team quickly discovered that their existing dashboards could not answer a simple question: was AI actually improving productivity?

They discuss how Mercari built new visibility into AI usage and software delivery, the bottlenecks they uncovered across the SDLC, why faster coding did not automatically translate into faster delivery, and the lessons they learned rolling out AI at scale. They also share how Mercari is rethinking software development around agents, feedback loops, and new ways of working.

Some takeaways:

Measuring AI impact

AI adoption alone does not guarantee business value. Mercari found that while AI usage increased rapidly across the organization, existing dashboards could not answer the leadership team’s most important question: whether AI was actually improving productivity.
Local optimization does not necessarily improve system-wide performance. Engineers reported working faster with AI tools, but end-to-end delivery metrics remained largely unchanged because bottlenecks elsewhere in the software delivery process continued to slow teams down.
Organizations need visibility into both AI usage and delivery outcomes. Mercari built new dashboards that combined AI tool data with SDLC metrics to better understand adoption, throughput, quality, and operational performance.

The reality of becoming AI-Native

AI adoption required a cultural transformation, not just a tooling rollout. Mercari’s CEO mandated company-wide AI adoption, but success depended on changing workflows, habits, and expectations across engineering, finance, legal, customer support, and other functions.
Different teams required different forms of enablement. Employees varied significantly in their technical backgrounds and familiarity with AI tools, making education, workshops, and support systems essential to driving adoption.
The goal was to rethink work itself. Rather than layering AI onto existing processes, Mercari challenged teams to reconsider what they built, how they built it, and how people worked together.

The bottlenecks AI exposed

AI revealed problems that already existed inside the organization. Review queues, CI instability, deployment friction, and support requests became more visible as coding accelerated.
Code generation was not the primary constraint. Engineers often spent more time waiting for approvals, navigating organizational boundaries, and dealing with infrastructure limitations than writing code.
System complexity amplified AI-related challenges. As AI-generated changes increased, existing architectural complexity and fragile workflows became harder to ignore.

Finding AI workflow opportunities

Mercari mapped AI opportunities across 33 domains. The AI task force reviewed functional areas across the company to identify where AI could automate work, where it could not, and where the strongest leverage points existed.
The biggest opportunities extended far beyond engineering. Role-specific workshops helped teams in finance, legal, design, operations, customer support, and other departments find practical AI use cases in their own workflows.
Early wins created proof points for broader adoption. Mercari saw measurable impact from support bots, accounting workflows, platform support automation, and Socrates, an internal BI agent that made company data easier to query and use.

Rethinking software development

Faster coding shifted attention upstream. As implementation became easier, planning, specification, and decision-making emerged as larger constraints on delivery speed.
Agent Spec-Driven Development moves AI earlier in the lifecycle. Mercari began using agents to analyze documentation, code, and organizational knowledge before implementation work started.
Future workflows will focus more on intent than execution. Teams increasingly define goals, constraints, and success criteria while agents handle larger portions of implementation and validation.

Preparing for an agent-driven future

Feedback loops matter more than ever. Mercari’s multi-loop SDLC emphasizes rapid validation, iterative learning, and increasingly autonomous agent workflows.
Behavioral change remains harder than technological change. Organizations must rethink ownership, accountability, and trust before they can fully benefit from agent-based development.
The path to AI-native development is iterative. Mercari expects continued setbacks and learning cycles, applying what they describe as the Stockdale Paradox: maintaining confidence in the destination while remaining honest about current challenges.