CDO Matters Podcast CDO Matters Ep. 94 | Rethinking Data Governance for AI
24 snips
Feb 7, 2026 Sarah Levy, CEO and founder of Uno, brings two decades of startup leadership across cybersecurity, healthcare, and fintech. She talks about why traditional data governance breaks under AI. They cover automated context and metadata, column-level lineage, semantic tensions and multiple truths. She outlines adaptive, probabilistic governance and practical starting points like lineage and business champions.
AI Snips
Chapters
Books
Transcript
Episode notes
Real-Time Context Beats Static Catalogs
- AI agents need real-time contextual metadata, not static catalogs, to produce trustworthy answers.
- Sarah Levy says stitching lineage, usage, ownership and docs into live context fixes hallucination risks.
Automate And Prioritize Certified Assets
- Automate metadata collection and classification across the whole stack and feed agents with prioritized, certified assets.
- Levy advises surfacing the two most relevant certified definitions to guide agent queries and build trust.
Centralize Metadata With Hybrid Storage
- Centralizing metadata is necessary but the backend must be hybrid for cost and performance.
- Sarah Levy notes heavy graph storage is expensive so use multiple databases with graph for relationships.
