Drawn from a real system migration affecting millions of users, this session walks through what AI agents for software delivery got right and what slipped through. The high volume mechanical work was the easy part, handled cleanly and exactly as expected. The harder problem showed up later: a failure that looked like success because telemetry couldn't tell the difference, and a scope gap that took an audit to find.
For financial services engineering and DevOps leaders, this setup will feel familiar. Your systems share the same conditions that made this migration hard: zero tolerance for outages, undocumented edge cases accumulated over years, and pressure to modernize without introducing new risk. Faster code generation is not faster software delivery. Join us to hear how GitLab approached it and what we learned.
You'll walk away knowing how to:
- Apply AI software delivery agents to migration projects
- Understand what worked, what didn't, and why
- Identify and close gaps before they reach users