This panel discussion explores what changes in data management when AI moves from analytics into operations, and when unstructured data becomes core input.

What the panel highlights

• AI changes what “important data” looks like. Claims, payouts, and operational workflows increasingly depend on documents, images, and text.
• Operational AI raises the bar for governance. Systems cannot be “down for a day,” and ownership must be shared with business units.
• Observability + iteration are practical foundations. Start from clear business goals, build tracing/monitoring, and keep humans in the loop early.
• Unstructured data creates new quality questions. Do we have the right data, is it correct and current, and is it presented in the right way (format, vectorization, etc.)?
• Big-bang “connect the LLM to everything” is risky. Start with targeted document sets, clean and govern content first, then expand.

Key takeaway

Making data ready for AI is not a side project. It changes risk, ownership, and quality requirements.

(If you want the panel recording embedded here too, I can try, but many panel videos are stored as embed blocks and may not paste cleanly.)