Columbia DAP Lab × Purple Fabric · 2025
What it takes to make AI agents enterprise-ready
Eugene Wu, Columbia University DAP Lab, in conversation with Aditya Rajashekaran
Eugene Wu is Associate Professor at Columbia University and Co-Director of the DAP Lab. In this session, he and Aditya Rajashekaran, AVP Product at Purple Fabric, work through the realities of deploying AI agents in production: where the biggest risks emerge, why building an agent is often easier than validating it, and what enterprises actually need before they can trust AI in regulated, high-stakes environments.
The discussion covers:
- the difference between agent development and traditional software development
- benchmarking, evaluation, and the challenge of proving reliability
- human-in-the-loop review, accountability, and trust
- data lineage, citations, traceability, and auditability
- why "gold standard" data can still be flawed
- the infrastructure and controls needed for scalable agentic systems
The examples come from insurance, wealth management and ESG work, and the conversation stays on the practical questions: operations, governance and system design.