Farah Serrano

Most AI projects fail because enterprises built data systems for humans, not agents.

Joined data creates new risks that field-level controls cannot detect or prevent.

Data governance designed for humans won't stop AI systems from failing on bad data.

Organizations can govern AI data without migrating it to a central warehouse.

Agents need catalogs that enforce semantics and permissions at query time, not just document data.

Static field labels miss sensitivity that emerges when AI agents combine data across queries.

AI systems need data structured for agents, not analysts—and they're fundamentally different.