← All speakers

Bio, Work & Ideas

Bruno Passos

Conference affiliation: booking.com · 2025

Bruno Passos is a Booking.com product leader specializing in developer experience and AI enablement and the creator of Cornr, a motorcycle-riding application built with generative AI. He focuses on whether coding assistants actually improve software development—and what engineering organizations must change to make those improvements measurable.

With more than 15 years in technology, Passos has led product work spanning developer experience and generative AI at Booking.com, whose engineering organization includes more than 3,000 developers. Years of experimentation had left its codebase crowded with obsolete feature flags and accumulated technical debt, increasing the time engineers spent navigating and maintaining existing systems.

When Booking.com introduced Sourcegraph’s Cody assistant, access alone failed to produce sustained adoption. Passos expanded model choice and codebase context while organizing workshops and hands-on experimentation. His analysis of coding-assistant outcomes associated daily Cody usage with approximately 30 percent more merge requests and smaller changes, while acknowledging that the methodology remained immature. Those findings describe an observed correlation, not established causation.

  • Evidence-based AI adoption: Passos evaluates assistants through lead time, review activity, change throughput, software quality and modernization outcomes instead of relying on speculative hours-saved estimates. He treats sustained usage and developer education as prerequisites for useful measurement.
  • Context-aware legacy-code modernization: His enterprise work combines code search, organizational context and structured prompts to identify obsolete feature flags, map migration work and tackle aging systems incrementally. His AI Engineer appearance alongside Beyang Liu describes how practical training helped engineers apply these tools to their own codebases.
  • AI-assisted product development: Cornr helps motorcyclists discover routes, record rides and connect with other riders.

Read the topics behind these talks

1 conference talk

Key ideas

Scroll to read ↓

Booking.com’s work with Sourcegraph connects AI adoption to delivery metrics, then tackles GraphQL generation, legacy migration, and code review through carefully selected context and explicit rules.

  • What did adopting AI actually change?
    0:51 ↗
  • Experimentation creates a cleanup problem
    1:51 ↗
  • Search first, then an assistant people can use
    5:17 ↗
  • Measure delivery, quality, and modernization
    8:30 ↗
  • Prompt customization becomes agent development
    11:49 ↗
  • Retrieve schema paths before generating GraphQL
    12:36 ↗
  • Scope the migration before attempting the rewrite
    13:54 ↗
  • Make code review specific and selective
    15:50 ↗
  • Declare service rules and enforce them earlier
    17:16 ↗
  • Education turns access into regular practice
    19:38 ↗

References