Review AI Agent

An AI agent is a software system that uses a model to choose and carry out multiple steps toward a goal, often by observing state, using tools, and adapting its next action.

Instead of answering once, an agent may plan, call software tools, inspect results, update its working state, and continue until it reaches a stopping condition or requires human input.

Multi-step action expands both usefulness and risk. Teams need explicit permissions, boundaries, checkpoints, observability, ownership, and reliable stopping and rollback paths.

An agent reviews a support queue, drafts responses from approved sources, creates proposed ticket updates, and pauses for a person before anything is sent.

Agent does not mean fully autonomous, conscious, or trustworthy. Many useful agents operate within narrow tasks and frequent human gates.

A chatbot answers; an agent can continue and act.

* [[actor|Actor]] * [[orchestration|Orchestration]] * [[objective-escape|Objective Escape]] * [[human-ownership|Human Ownership]] * [[state-transition-protocol|State Transition Protocol]]

* [NIST NCCoE, *Accelerating the Adoption of Software and Artificial Intelligence Agent Identity and Authorization*](https://csrc.nist.gov/pubs/other/2026/02/05/accelerating-the-adoption-of-software-and-ai-agent/ipd) — official concept paper defining AI agents and examining their authority, identity, and access risks.

* Define action boundaries with [[human-ownership|Human Ownership]] and [[state-transition-protocol|State Transition Protocol]].