Agentic AI
Agentic AI describes AI systems that pursue goals by taking actions in real systems — planning, using tools and adapting — rather than only generating text.
Agentic AI is the shift from AI that produces output to AI that produces outcomes. Instead of answering a question and stopping, an agentic system is given a goal, a set of tools and a set of boundaries, then plans a sequence of steps, calls the tools, checks the results and adapts.
In a business context that usually means an AI system with its own identity and permissions inside your software — able to read records, invoke business logic and complete a task, with its actions recorded for audit. Microsoft's Dynamics 365 agents follow exactly this pattern: each runs under a dedicated agent user identity with its own security roles and an activity log.
The practical implication is that agentic AI is a governance question as much as a technology one. Once software can act on your behalf, the important decisions are what it's allowed to do, who owns it, and how you would prove what it did.