Data Contracts Are Good for AI
summary
Argues that standardized data contracts (ODCS v3) turn metadata into an executable source of truth and cut AI risk by enforcing schemas, SLAs and data‑quality tolerances; shows a YAML contract model, a data contract CLI validator and an ecosystem that grew from 5 to 35 contributors across ~30 companies after a PayPal template was open‑sourced and incubated; documents 12 SLIs (example: a 4‑day latency SLI), seven data‑quality criteria, and a live validation demo that fails fast on schema and quality breaks.
AI-written summary · about
chapters (14)
- 0:00Introduction and context
- 1:34Understanding data contracts
- 4:32The role of data contracts
- 8:44Standardization efforts
- 10:56Building the ecosystem
- 14:55Contract structure and design
- 16:32Key components of contracts
- 22:27Implementation examples
- 28:03Service levels and quality
- 28:43Data contracts for AI
- 30:34Standardizing the AI ladder
- 31:34Demonstrating the CLI tool
- 33:07Conclusion and takeaways
- 35:39Q&A and closing
related talks
50mData Contracts VS Data Chaos: A Survival Story
27mLive in the Data Wild West: The Data Contracts Sheriff
38mRedefining Test Data Strategy in the Gen AI Era
47mStructured Outputs in OpenAI: Building Reliable JSON-Based AI Integrations
25mYou Are What You Ingest: Why Data Quality Matters to Smart Contracts
37mSponsored Session: Trust by Design: Engineering Reliable Enterprise AI Workflows with Open Source
46mEsther Huyer & Raymonde Weyzen
41mData Quality: The Holy Grail For A Data Fluent Organization