About this session:
Join us for a workflow demo built for data scientists adding AI to workflows where the results have to hold up, whether to a reviewer, a security team, or a regulator. We'll start by diving into where MCP fits and why it offers more control than skills for reproducible workflows, then go deep on two real servers: a CDISC conformance validator that wraps an existing rules engine, and another that uses a pinned, separate LLM to classify adverse events.
From there, we'll show where that same pattern extends further. The use case comes from clinical research, but the architecture doesn't. Any workflow with a rules-based check against a standard, or free text mapped to a controlled taxonomy, is built the same way.
Beyond showing you how each server works, our speakers will walk you through the practical value each one delivers: what it means when you need to reproduce a result, what it saves your team from stitching together by hand, and what questions it lets you answer the next time someone asks how AI contributed to the analysis.
Bring your questions for a live Q&A at the end.
📅 Wednesday, September 30, 2026 🕐 11:00 AM EDT
💻 Live Demo + Q&A with Samiul and Matt
