What challenges did Wayfair face before AI implementation?

Post-commit CI build failures were slowing developer velocity because engineers had to manually analyze logs and repeat fixes. The CI/CD pipeline became a bottleneck at high scale, with inefficiency across more than 25,000 builds per day.

What was the impact of the AI solution?

58% reduction in MTTR, from 26 minutes to 11 minutes. 12,000+ build retries avoided monthly. CI fix time cut from 30 minutes to under 5 minutes. 80% reduction in context switching. Over 31,000 engineering hours are projected to be saved annually. Developer NPS increased from 4.8 to 8.2.

Source

GCP

Wayfair: Gemini-powered AI CI/CD intelligence for faster build-failure remediation

Use case typeDeveloper productivityUpdated 2 days ago

Wayfair built a GenAI-powered CI/CD intelligence system to reduce developer toil caused by post-commit build failures. The system combines Google Cloud's Gemini model with a custom RAG pipeline over Buildkite logs, MCP metadata, and historical failure data to generate explanations and fix recommendations in Slack and inside developer IDEs. The solution is live in production and used by about 70% of Wayfair developers.

Organization: Wayfair
Industry: Tech & Comms
Location: United States
Published: July 2026

Reported outcomes

−80%

context switchingOther quantified impact

−58%Mean time to recovery12,000 countbuild retries avoided monthly−83.3%CI fix time31,000 hoursengineering hours saved annually70.8 pointsdeveloper NPS

Strategic outcomes

Scale & capacityImproved developer productivity and satisfactionScale & capacitySupported high-volume engineering operations at scale

Primary read

Use case focus

Showing 2 of 2

1Developer productivity
2Workflow automation

Post-commit CI build failures were slowing developer velocity because engineers had to manually analyze logs and repeat fixes.
The CI/CD pipeline became a bottleneck at high scale, with inefficiency across more than 25,000 builds per day.

Wayfair built an AI-powered build failure remediation system using Gemini and Gemini Enterprise Agent Platform.
A custom RAG pipeline combines Buildkite logs, MCP metadata, and historical failure data to generate LLM explanations and recommended fixes in Slack.
An MCP server and secure agentic IDE integration let developers access log insights and fix suggestions directly in tools such as Cursor.
A feedback loop continuously refines recommendations based on developer input.

Technologies

Gemini Gemini Enterprise Agent Platform Buildkite MCP server Cursor Slack RAG

58% reduction in MTTR, from 26 minutes to 11 minutes.
12,000+ build retries avoided monthly.
CI fix time cut from 30 minutes to under 5 minutes.
80% reduction in context switching.
Over 31,000 engineering hours are projected to be saved annually.
Developer NPS increased from 4.8 to 8.2.

Architecture

The solution uses Google Cloud's Gemini model and Gemini Enterprise Agent Platform, a custom RAG pipeline over Buildkite logs, MCP metadata, and historical failure data, Slack delivery for recommendations, and MCP server-based secure agentic IDE access for tools such as Cursor, with a feedback loop to refine recommendations.

Sources & evidence1

Groundedness: 5/5Type: Customer StoryPublished: Jul 11, 2026Publisher: Google CloudEvidence: PrimaryConfidence: High

Primary source

AI-generated summary. Verify important details with the linked sources before relying on this case.

Explore related AI use cases

Customer evidence