Anurag Sahay

Most companies have shipped AI by now: copilots, dashboards, recommendation engines. The models work, the data platforms exist, yet AI still usually sits behind a human. It advises, but rarely acts. That gap comes down to orchestration, governance and organization design. This session will explore what must be in place before AI can run autonomously, and what changes when it does. We’ll walk through enterprise scenarios where agentic AI runs on a Databricks backbone, and the architecture and governance choices that made it possible. We’ll also cover where human oversight still belongs and it's importance. Highlights: - Why AI programs get stuck between "working demo" and "running in production," and what closes that gap - Walkthrough of agentic workflows on Databricks including decision movement between AI and people - Nagarro's Fluidic Intelligence framework for treating AI as a continuous system - What leadership teams need to start doing as well as stop doing