glossary term · cross business

AI agent

Learn the meaning of an AI agent, see a synthetic support example, and distinguish model-directed tool choices from a fixed AI-assisted workflow.

Author: Founder identity pending confirmation
Published: Pending first public release · Fact-check: Human review pending
beginner · understand · Approximately 2 minutes to read
Evidence: editorial · Maintenance: current

Definition

An AI agent is a software system that uses a model to choose steps toward a goal, within defined tools and permissions. It can use the result of one step to decide what to do next, including stopping or asking for help. Its choices need boundaries and testing. A workflow that always follows the same route is not necessarily an agent, even when one step uses AI.

Owner example

In a synthetic support example, the system decides whether an approved FAQ lookup or a read-only order lookup could answer the question. If neither provides enough evidence, it stops and asks a person. It cannot send a reply or change an order. This describes a design, not an executed test.

Non-example and limitation

A fixed sequence that reads an email, asks a model to categorize it, and adds a review row is an AI-assisted workflow. The model interprets language, but it does not choose the next tool. Calling the sequence an agent would hide that distinction.

Model-directed choice does not guarantee correctness or justify broader permissions. More possible paths mean more behavior to test.

Related terms

Tool calling means requesting a defined software capability. Human approval means a person authorizes an action before it happens. AI workflow is the broader term for a process that includes an AI step. Separate term pages for these concepts are planned; they are not yet published.

Deeper guide and practical link

Read AI agents explained for business owners for the same task compared across manual, assisted, and agentic handling. Then inspect the support triage playbook as a practical fixed workflow with a human review boundary.

Sources

The workflow/agent distinction follows the design framing in Anthropic’s Building effective agents. This entry is an editorial explanation, not a claim of tested capability.