How Wayfair’s AI Solution Transformed Developer Feedback Loops

Join us with Pushkar Sharan from Wayfair, recent winner of the prestigious DORA award for Embracing Artificial Intelligence, as he shares how his Developer Experience team leveraged Buildkite’s MCP to transform build failure remediation, pushing the boundaries of what a developer feedback loop can look like.

Facing 25,000+ builds per day, where even small failures translate into thousands of lost developer hours, Wayfair's solution combines Retrieval-Augmented Generation (RAG) with Buildkite's MCP integration to eliminate productivity drain. This combination became the bridge to push CI insights directly into the developer workflow IDE, allowing developers to receive AI-powered fix recommendations without leaving their coding environment, eliminating costly context switching and capturing organizational knowledge.

The results speak for themselves: Mean Time to Resolution (MTTR) dropped from 26 to 10 minutes (a 58% improvement), 12,000+ unnecessary build retries avoided monthly, and developer NPS scores jumping from 4.8 to 8.2. Most impressively, the solution is projected to save 30,000+ engineering hours annually at full adoption.

In this discussion, we'll cover:

• The solution that earned Wayfair the DORA award for Embracing Artificial Intelligence
• How Wayfair built a RAG system trained on 2.5+ million past Buildkite build failures and fixes
• The implementation details of how Buildkite annotations surface AI recommendations via MCP
• The technical architecture enabling developers to trigger Buildkite builds from their IDE and receive fix recommendations without context switching
• Security considerations when connecting Buildkite with AI services via OAuth-based token exchange
• Real-world examples of complex Buildkite build failures automatically remediated through MCP

Discover how Wayfair transformed their Buildkite pipeline from a potential productivity bottleneck into a developer experience advantage, and how your team can implement similar AI-powered improvements to your CI/CD workflow.