The AI in Business Podcast How Open Context Layers Help Enterprises Build, Govern, & Scale Agentic AI - with Prukalpa Sankar of Atlan
12 snips
Jan 8, 2026 
Prukalpa Sankar, Co-founder and CEO of Atlan, leads a platform for data and AI governance. In their conversation, he reveals why 75-95% of enterprise AI pilots fail, attributing it to context gaps rather than just data complexity. He discusses the importance of a dynamic context layer that integrates data, business meaning, and governance. Prukalpa also shares insights on improving accuracy and ROI through context-ready practices, and emphasizes that businesses must co-own AI adoption to remain relevant in the rapidly evolving landscape.
AI Snips
Chapters
Transcript
Episode notes
Context Is The Core Failure Point
- AI projects fail not primarily from bad models but from missing machine-readable business context.
- Embedding shared meaning into data is essential for reliable production AI.
Top 10 Customers Revealed Context Gaps
- A customer rolled out an AI data analyst and found ambiguity in "top 10 customers" definitions across teams.
- The agent needed layered context (revenue vs. NPS) to answer correctly and learn from interactions.
Make Context Dynamic And Self-Updating
- Build a dynamic context layer that updates with business interactions and feedback loops.
- Capture live user decisions and feed them back into the context store to improve agent behavior.