Piyush Ahuja

What does data governance look like when your biggest data consumer doesn't sleep, doesn't follow SOPs, and makes thousands of decisions every second? When a majority of your users bypass official AI tools entirely? When a single ungoverned prompt can leak training data, customer PII, or regulated information across your entire enterprise? These aren't hypothetical questions — they're the operational reality of GenAI in 2026. And the governance frameworks most enterprises rely on were built for an era when data consumers were human, deterministic, and slow. This talk presents DataGovOps: governance as code, dynamic access policies, AI-assisted classification, and end-to-end lineage from source through agent output. We'll explore why legacy governance fails under agentic workloads, how to architect consent and access as programmable policy, the recursion problem of governing AI with AI, and practical patterns for staying compliant under regimes like GDPR & DPDPA — without strangling the velocity that makes AI worth deploying in the first place.