Before You Give AI Agents a Map, You Have to Build the City

AI agents can reason, plan, search for information, and use tools. But intelligence alone is not enough. To make good decisions, an agent must understand the enterprise around it: what business terms mean, how people, data, applications, and systems are connected, where relevant information is stored, what is happening at that moment, and which actions are permitted.
In this talk, we will use a city-building analogy to explain how organizations can prepare their enterprises for reliable AI agents. Taxonomy creates a shared language, ontology defines meaning and relationships, knowledge graphs connect enterprise information, RAG helps retrieve relevant knowledge, and context graphs provide current information such as identity, risk, approvals, time, location, and system state.
The session will also explain why being able to access a system does not automatically give an agent permission to perform every action. AI agents need both a clear map and appropriate rules, boundaries, and controls before they can operate safely.
The key message is simple: before giving AI agents more intelligence, organizations must build an enterprise environment that agents can understand, navigate, and trust.
Speaker(s): , Phani
Virtual: https://events.vtools.ieee.org/m/571282