Customer data glossary

AI agent

By Art Kukharenko · Published 2026-10-17 · Updated 2026-10-18

An AI agent is software that uses an AI model to pursue a goal by planning steps, calling tools and acting on systems, often with limited human intervention.

Part of the Learn hub and the customer data glossary.

What it means in practice

An agent differs from a chatbot in that it takes actions. A loop of model reasoning, tool calls and observation lets it break a goal into steps, query systems, draft or change things and check results. In customer data work, examples include agent-assisted segmentation (drafting a segment from a plain-language request), workflow QA (checking a journey before launch), anomaly detection and reporting.

There is no agreed legal definition of "agent". The EU AI Act regulates "AI systems", defined in Article 3(1) as machine-based systems designed to operate with varying levels of autonomy that infer from their input how to generate outputs such as predictions, content, recommendations or decisions that can influence physical or virtual environments 1. An agent will usually be an AI system under the Act; whether it is high-risk depends on its use, not on being an agent.

Controls are mostly old ones applied to a new actor: least-privilege tools, read-only by default, human approval for anything that sends, exports or deletes, minimal context (data minimisation), spend and rate limits, and a complete audit ledger.

Example (health insurer): an agent proposes a segment definition from a plain-language request and a human approves it. The agent cannot export member lists, and each step is logged.

Model Context Protocol (MCP): how agents reach tools.
Explainability: reasons for agent output.
Audit ledger: the record of agent actions.
Model lineage: which model acted.
High-risk AI system: when use triggers regulation.

Where it matters

Running AI agents on customer data without leaving your boundary: architecture patterns.
Seven things AI agents should never do with customer data: and how to enforce it.

Back to the customer data glossary or the Learn hub.

Frequently asked questions

Is an AI agent the same as a large language model?

No. A model generates output. An agent wraps a model with tools, memory and a control loop so that it can act.

Are AI agents regulated?

Not as a category. They fall under the rules that cover their use: data protection law, sector rules and, in the EU, the AI Act for AI systems.

What is agent washing?

Gartner's term for relabelling chatbots or simple automation as agents. Ask vendors what the system decides, which tools it can call and who approves its actions.

Sources

Regulation (EU) 2024/1689 (AI Act), Article 3(1) Open source

This entry is general information, not legal advice.