Glossary

AI agent

An AI system (typically built on a large language model) that takes actions in the world via tools, follows instructions to accomplish a goal, and can operate with limited or no human intervention per step.

An AI agent is an AI system — usually built on a large language model like Claude or GPT — that can take actions in the world through tools (APIs, integrations, code execution), follow multi-step instructions, and make decisions about what to do next without a human re-prompting at every step.

The distinguishing properties of an AI agent vs other AI usage:

  1. Tool use. The agent can call APIs, query databases, send emails, update CRMs — not just produce text.
  2. Multi-step reasoning. The agent can break a complex task into sub-steps and execute them in sequence.
  3. State + memory. The agent maintains context across steps (and often across sessions).
  4. Autonomy with guardrails. Within defined boundaries, the agent acts without human approval at every step.
In business automation, AI agents typically handle structured workflows — lead routing, support triage, document drafting, billing ops — where the steps are predictable enough to specify in a prompt but variable enough to require reasoning at each step.

Example

A sales agent receives a lead webhook → calls the Apollo API to enrich the contact → queries the company CRM to check if this is an existing account → scores the lead against the company's ICP using a structured rubric → posts the result to Slack with a routing recommendation. All of that runs in 20 seconds without a human in the loop.

Related terms

Apply this

Get a real automation audit on your business.

Pay $4,500, fill an intake, get a Claude Opus-written strategic roadmap inside a week.

Start the audit →