AI Agents and the Data Lake (w/ Lauren Anderson)

Podcast episode. Lauren Anderson, head of enterprise data platform at Okta, discusses how identity sits at the center of two shifts: AI agents and the open data lake.

Curated Topics

• Agent access governance: Agents should query centralized, agent-ready stores with governance run once — policies, roles, tracking on a central plane
• Semantic layer as essential bridge: Creating semantic views must get easier and more automated; semantics should inform policy application
• Row-level security for agents: Conversational intelligence data must be constrained to what the requesting user can access; aggregations can be broader with anonymization
• Analytics engineer vs data engineer roles in agent era: AEs should own semantics (tooling, vendor choices, shared business language); DEs should optimize for consistency, scale, and governance
• MCP experimentation: Okta is early-stage, building support and insight use cases; nothing broad in production yet
• Rethinking relational assumptions: Many initial agent questions are intentionally simple — speed and reasonable accuracy trump perfect sophistication

RDCO-Relevant Links

• Semantic layer governance for agent access — directly relevant to dbt consulting positioning
• MCP as agent-data interface — validates our agent-ready data infrastructure work
• Central governance plane concept — maps to SDG pipeline governance patterns