Glossary
Updated
Agentic AI represents the shift from "passive" AI (which generates text or insights) to "active" AI (which interacts with the world). Unlike a standard Large Language Model (LLM) that waits for a prompt to produce an output, an Agentic system has agency: it can reason, plan, use tools (like APIs or browsers), and execute a chain of tasks to achieve an objective. This autonomy introduces new governance challenges, as the risk shifts from what the AI says (content) to what the AI does (actions).
Example
A travel booking company moves from a Generative AI chatbot (which only suggests hotels) to an Agentic AI system. The Agentic system doesn't just suggest the hotel; it autonomously checks the user's calendar, negotiates the best rate via an API, books the room, and charges the corporate credit card, all without the user clicking a button.
