Foundations

What is an AI agent?

An AI agent is software that uses a model to act — holding credentials, calling tools, completing work. What counts as an agent, the types you will meet, and the moment one becomes your operational responsibility.

Strip away the marketing and an AI agent is three things bolted together: a model that decides, tools that act, and a loop that keeps deciding and acting until a goal is met. The model alone is not an agent — it only produces text. The automation alone is not an agent — it only follows rules. The agent is the combination: software that can be given an outcome rather than a procedure, and that holds real credentials to real systems while it pursues it.

That definition does useful work because the word is applied to almost everything right now. A scripted chatbot, a coding assistant, and an autonomous procurement workflow are all sold as "AI agents," and the label tells you nothing about which one can spend money. The questions that classify a system are operational: does it carry a goal across steps without re-prompting, does it write to anything, and how far does its loop run unattended? The agentic AI pillar covers the quality those questions measure; the agents-vs-chatbots and agents-vs-agentic comparisons draw the lines in detail.

The types you will actually meet

Classic AI theory sorts agents by sophistication — reflex agents that map condition to action, model-based agents that track state, goal-based and utility-based agents that plan. The taxonomy that matters operationally is simpler and crosses those lines: what does it do, and what can it touch. Coding agents open pull requests. Support agents read and write customer records. Voice agents make spoken commitments on calls. Back-office agents reconcile systems of record. Multi-agent systems split one workflow across several specialists. And no-code agents are any of the above, assembled by whoever had an afternoon and a license — which is why they dominate the unregistered population.

Whatever the type, the anatomy repeats: model, tools, state, orchestration — the architecture page walks it properly. The tool list, not the model, is the agent's real capability boundary, and that single fact organises most of what follows.

When an agent becomes your problem

An agent in production is an actor inside your organisation: it holds credentials, takes actions at machine speed, and works around the clock without supervision unless you built the supervision in. Readiness for that actor is concrete, sequenced work — an inventory of what already runs, an identity and scoped permissions per agent, security against the attacks agents specifically attract, governance that decides who may approve and widen them, and autonomy granted on evidence rather than enthusiasm. Teams meet this work either before their first consequential agent or during their first agent incident; the content of the work is identical, the price is not.

Where to go from here

If you are deciding whether and what to build, start with where agents pay in business and the build guide. If agents are already running around you, start with the inventory. And to place your organisation on the readiness journey, the maturity curve maps the stages — the assessment will tell you which one you are standing in.

For deeper coverage of specific agent types and deployment contexts, the explainer pages on chatbot agents, building AI agents, vertical agents, personal agents, small-business deployments, systems integration, and intelligent agents cover the specifics.

Frequently asked questions

What is an AI agent?

An AI agent is software that combines a model that decides, tools that act, and a loop that keeps deciding and acting until a goal is met. Unlike a model alone (which only produces text) or automation alone (which only follows rules), an agent can be given an outcome and holds real credentials to pursue it.

What is the difference between an AI agent and a chatbot?

A chatbot responds turn by turn and produces text for a person. An agent carries a goal across steps without re-prompting, calls tools, and can write to real systems. The operational test is whether it acts on its own and what it is allowed to touch.

What types of AI agents are there?

The taxonomy that matters operationally is "what does it do and what can it touch": coding agents that open pull requests, support agents that read and write customer records, voice agents that make commitments on calls, back-office agents that reconcile systems, multi-agent systems, and no-code agents assembled by anyone with a licence.

When does an AI agent become a security and governance concern?

The moment it runs in production. It then holds credentials, acts at machine speed, and works unsupervised unless you built the supervision in. Readiness is concrete work: inventory, a per-agent identity and scoped permissions, security against agent-specific attacks, governance, and autonomy granted on evidence.

How do I start managing the AI agents already running in my organisation?

Start with an inventory of what already exists — you cannot govern, secure, or assess what you have not listed. From there, assign named owners, move agents off personal tokens onto scoped identities, and add runtime visibility into the tools they call and the data they touch.

Is your organisation ready for AI agents?

Take the assessment →