Stop Reviewing Schemas: How Intuit Made Developers Faster by Automating Governance
summary
Intuit replaces slow manual schema reviews with a CI/CD gate, dev portal and IDE plugin that run a custom linter, an AI-powered schema analysis (LLM prompts), and semantic collision detection using cosine similarity to compare descriptions. The system auto-fails pipelines on rule errors, supports global and per-subgraph exceptions, and surfaces high-quality fix guidance; it performs about 50,000 schema validations/month and drops onboarding from ~90+ days to under 10 days while enabling future auto-approvals and AI fix suggestions.
AI-written summary · about
chapters (7)
- 0:00Automated schema governance intro
- 1:06Manual reviews fail at scale
- 2:49Smart tools for governance
- 8:49Demo: schema onboarding
- 14:29Impact on developer velocity
- 17:22Key lessons learned
- 20:56Future of self-governing APIs
related talks
29mBold Move: Automating Away Developer Friction
05mLint All The Things
09mLightning Talk: How Intuit Enables GitOps at Scale For All Its Developers
29mSchemas: The Key To Data Happiness
30mFailing Fast with Data: Bringing CI to the Land of Data
12mShared Schema Policies and Automatic Standards Across Your Company’s Teams
01h 02mEffective Code Review
26mYour pipeline is the bottleneck