AI developers are building some of the most exciting technology of our time—new agents, new workflows, new assistants, new automation layers. But the moment you try to bring any of this into a large enterprise, you run headfirst into realities that aren’t obvious from the outside: decades of legacy systems, thousands of long-tail applications, fragmented identity boundaries, inconsistent APIs, compliance constraints, and user environments shaped entirely around humans.

This session is a practical guide to that world, from someone who's spent 30+ years in enterprise IT. We’ll break down what AI builders consistently underestimate (and overestimate), why great demos fail in real customer environments, and what you actually need to design for if you want your AI product, agent, or platform to gain enterprise traction. You’ll learn why GUI-level automation isn’t a hack, why “just use the API” rarely works at scale, why enterprises adopt new tech slowly and nonlinearly, and how to bridge your modern AI ideas with the reality of old infrastructure.

Whether you’re building agents, copilots, devtools, enterprise integrations, or AI-native products, this session gives you the mental model you need to unlock one of AI’s biggest unrealized TAMs: applying modern AI to the messy, high-value workflows of large organizations.