Speed Without Sacrifice: How Wayfair Transforms DevEx With AI and MCP
summary
Wayfair embeds AI across the GraphQL lifecycle to cut review and testing friction: GraphQL Schema Copilot uses RAG with embeddings over 12 MB of internal docs to post inline suggestions and apply one‑click patches, cutting time‑to‑first‑comment from 2.5 days to 0 and reviews from 3 days to 1.5 (≈$150k saved) at about half a cent per review, with Gemini Flash hitting an 83% pass rate. Parallelly, AI response mocking stitches production data with LLM‑generated fields validated by a Z schema, speeding time‑to‑market ~15% (> $100k/year).
AI-written summary · about
chapters (9)
- 0:00Introduction to talk
- 1:03Hidden friction in SDLC
- 2:08AI for faster schema reviews
- 6:05Building GraphQL Schema Copilot
- 11:49Testing bottleneck moves downstream
- 12:31AI-powered GraphQL response mocking
- 16:46Mocking service architecture
- 22:49Key learnings and takeaways
- 23:32Summary and Q&A
related talks
33mSpeed Without Sacrifice: How Wayfair Transforms DevEx with AI and MCP
46mPanel: AI in the Loop - Left-Shifting Platform DX and Governance
26mEvolving GraphQL: Designing APIs That Deliver Insights with Intelligent Resolv…
47mSchema Design in the Age of AI: Pragmatic Patterns to Evolve and Ship Fast
29mClosing the Loop: How GraphQL Gives Coding Agents Eyes on What Actually Matters
19mThe AI Code Review Bottleneck: Escaping the Velocity Trap
27mPanel: APIs for AIs
26mAsk, Don't Fetch: Why GraphQL + MCP Is the Native Language of AI Agents