Panel: AI in the Loop - Left-Shifting Platform DX and Governance
summary
Apollo MCP v1.0 and Agentic power concrete tools: an Intuit IDE plugin runs AI linting and collision detection on schemas during CI to catch semantic collisions and link fixes to docs; Wayfair replaces first human review so time to first comment drops from 3.5 days to zero and approval time falls from 6 to ~4.5 days saving ~$200k in six months; Expedia queries DataDog via MCP to audit 90+ subgraphs, cut JVM instances and report up to 50% cloud cost reduction; Capital One uses RAG grounding, human-in-loop approvals, and LLM-generated mock data to move releases from weeks toward hours.
AI-written summary · about
chapters (14)
- 0:00Introduction
- 1:06First agent interactions
- 6:00Left shifting with GenAI
- 7:06Governance at Intuit
- 13:27Testing with mock data
- 15:15Agentic workflows at Capital One
- 21:17GenAI over scripts
- 23:01Schema automation at Wayfair
- 27:14Observability at Expedia
- 32:45Prompts and automation
- 35:13Scaling MCP challenges
- 38:55Hallucinations and validation
- 43:22Updating governance rules
- 45:08Call to action
related talks
27mPanel: APIs for AIs
50mPanel Software Engineering in the age of AI
44mPanel Discussion | The AI Revolution in Testing: Opportunities and Challenges
48mMastering the AI Shift: Adaptive Strategies for Software Evolution
50mPanel: Beyond The Vibe: Code Quality First
31mModern Architecture for AI-Augmented Development Teams
25mPanel The State of Web Development in 2026
28mSpeed Without Sacrifice: How Wayfair Transforms DevEx With AI and MCP