All glossary terms
A Foundations Agentic AI

AI agent

An AI agent is software that takes a goal and decides its own steps to reach it. The distinguishing feature is not intelligence. It is discretion. If a person wrote down the sequence in advance, what you have is automation with AI inside it.

Definition

An AI agent is a software system that pursues a goal by deciding its own next step. It perceives its context, plans, calls tools or other systems to act, and observes the result before deciding again. What separates an agent from automation is that the sequence of steps is chosen at runtime rather than written in advance.

What is an AI agent?

An AI agent is software that takes a goal and figures out its own steps to reach it, using a model to reason and tools to act. What sets it apart isn't how intelligent it is. It's that nobody wrote down the sequence it follows.

Start with the plainest version. Traditional software does what it was told, in the order it was told. You describe the steps, it performs them, and if it meets something the steps do not cover, it stops or fails. An AI agent is given an objective instead of instructions. It works out the steps itself, and it can handle situations nobody anticipated because it was never following a fixed list.

A short example makes the difference concrete. Asked to process an invoice, conventional automation follows a defined path: read these fields, check them against this table, route to that queue. An agent is told to get the invoice paid correctly. It might read the invoice, notice the purchase order number is missing, search for a matching order, find a discrepancy in the quantity, and escalate to a person with the discrepancy explained. Nobody scripted that path. It was assembled in response to what the agent found.

Why nobody agrees on the definition

The term is used inconsistently, and it is worth knowing why before evaluating anything sold as one. There is no standards body for what qualifies, so in practice vendors set the threshold wherever their product happens to sit. A chatbot that can call one function gets marketed as an agent. So does a system that runs unattended for hours across a dozen systems. Both claims use the same word for capabilities orders of magnitude apart.

That makes "is it an agent" a poor question and "how much does it decide" a good one. The useful reading is a range rather than a category: at one end, software that follows a fixed path with a model doing one step inside it; at the other, software that determines its own path, selects its own tools, and decides when it is finished. Most things being sold sit closer to the first end than their marketing implies, which is the subject of agent washing.

What differs an AI Agent from Automation?

One question settles it in most cases: who decided the order of the steps? If a developer or a business analyst wrote the sequence, it is automation, however much AI sits inside individual steps. If the software determines the sequence while running, it is an agent.

Three follow-on properties usually travel with that answer, and they are worth checking because they are what make an agent operationally different rather than just architecturally different.

  • It can take actions, not just produce output. A system that returns a recommendation for a person to act on is doing something meaningfully different from one that issues the credit itself. The second changes what happens when it is wrong.
  • It handles cases nobody enumerated. Automation meeting an unlisted condition stops. An agent attempts something. That is the benefit and the risk in one sentence.
  • It decides when it is done. This is the property most often absent, and the next section explains why its absence causes more production incidents than any other single gap.

Automation with AI in it is not a lesser thing. For a high-volume process with a known shape, a fixed path with a model handling one interpretive step is usually the better engineering choice: it is cheaper, faster, easier to test, and it can be approved once. The reason the distinction matters is not status. It is that the two need completely different governance, and treating an agent like automation is how organizations end up unable to explain what their software did.

What is an AI agent made of?

Five parts: a goal, a reasoning model, memory, tools, and a stopping condition. The first four appear in every vendor diagram. The fifth is routinely left out, and its absence is behind most runaway cost and looping failures in production.

The components below describe what has to be present for an agent to function. How they are implemented varies enormously between frameworks and platforms, but the parts themselves are consistent.

Five components of an AI agent arranged around a central reasoning loop. The goal feeds in from the left. Memory and tools attach to the loop. The stopping condition sits at the exit and is highlighted in amber as the component most often missing. A caption notes that without a stopping condition an agent does not fail, it continues.
  1. 01
    A goalA stated objective, rather than a list of steps. How precisely it is scoped determines almost everything downstream, because a vague goal produces a wide search for ways to satisfy it. Failure mode: objectives broad enough to be technically satisfied in ways nobody wanted.
  2. 02
    A reasoning modelUsually a large language model, deciding what to do next given the goal and what has happened so far. This is the part people mean by "the AI," and it is one component of five. Failure mode: confident reasoning over a wrong premise.
  3. 03
    MemoryBoth short-term context within a single run and, in some designs, persistence across runs. Without it the agent forgets what it already tried. Failure mode: repeating an action already taken, which matters greatly if the action was a payment.
  4. 04
    ToolsThe systems, APIs, and data the agent can actually reach and act on. Tools are where an agent stops being a text generator and starts having consequences. Failure mode: broader access than the task requires, inherited rather than granted.
  5. 05
    A stopping conditionA definition of done, plus limits on steps, time, and spend, and a rule for when to hand off to a person. Failure mode: the agent does not fail, it continues, which is how a small task becomes a large invoice and a loop becomes an incident.

Component five deserves extra attention because of how quietly it fails. The other four fail in obvious ways: a bad tool call throws an error; a missing memory causes an obvious repeat. But an agent with no stopping condition just keeps going correctly, without ever wrapping up. So, the warning sign isn't an error alert, its rising cost, slower performance, and eventually someone having to step in manually. It's also the component least likely to show up in a demo, since demos end when the presenter decides to stop them.

What types of AI agents are there?

Two taxonomies are in circulation. The classical one from AI research grades agents by how they decide, from simple reflex through learning agents. The practical enterprise one groups them by the work they do. Both are useful, and they answer different questions.

The academic taxonomy is what turns up in textbooks and interviews. The functional grouping is what turns up in procurement. Knowing which one somebody is using saves a lot of confusion.

Classical agent types from AI research, what distinguishes each, and where each appears in enterprise use
Type How it decides Typical use Limitation
Reflex Reacts to the current input using fixed rules, with no memory Simple filtering, alerting, and routing Cannot handle anything requiring context
Model-based Keeps an internal picture of the world and reasons over it Monitoring where current state matters Only as good as the model of the world
Goal-based Plans a sequence of actions to reach a stated objective Most enterprise agents sit here Treats all successful paths as equal
Utility-based Weighs competing options against a measure of value Pricing, routing, resource allocation Requires the value measure to be right
Learning Adjusts behaviour based on the outcomes of past actions Personalization and long-running optimization Behaviour drifts, so it needs continuous evaluation

In practice, enterprise conversations use a functional grouping instead: agents are described by the work they do, such as customer service agents, procurement agents, reconciliation agents, or security triage agents. That framing is more useful for scoping and worse for architecture, because two agents doing very different technical things get the same label. When several agents work together on one outcome, that is multi-agent orchestration.

How is an AI agent different from a chatbot, a copilot, or RPA?

A chatbot has a conversation. A copilot assists a person who remains firmly in charge. RPA replays a recorded sequence exactly, step for step. An agent decides its own path and acts on it independently. The clearest way to tell them apart is this: does the software choose its own path, and is a human built into that loop by design, or left out of it?

These four are the most confused terms in the category, largely because products routinely sit between them and are marketed as whichever sounds strongest.

AI agent compared with chatbot, copilot, RPA, workflow automation, and assistant
Term What it does Who decides the steps Human in the loop
AI agent Pursues a goal and takes actions in systems The software, at runtime By exception, at defined checkpoints
Chatbot Answers questions in conversation Mostly a scripted or retrieval flow The user drives every turn
Copilot Suggests and drafts inside a person's work The person, who accepts or rejects Always, by design
RPA Repeats a recorded sequence of interface actions A developer, in advance Only when it breaks
Workflow Runs a defined process across systems A designer, in advance At designed approval points
Assistant Handles requests within a bounded scope Varies, often a mix Usually per request

Two of these distinctions carry real consequences. Copilot versus agent is a question of who is accountable: a copilot's output passes through a person who reviews it, which is a control, whereas an agent acting directly removes that control and has to replace it with something else. And RPA versus agent is a question of brittleness against unpredictability: RPA does exactly the same thing every time and breaks when a screen changes, while an agent adapts and is correspondingly harder to predict. Neither property is better in the abstract. They suit different problems.

What goes wrong with AI agents in production?

Five failures recur, and none of them is the model being unintelligent. Unbounded runs, wrong premises, inherited permissions, no owner, and no record. Each is a property of how the agent was deployed rather than of how well it reasons.

This is the part of the subject that tends to be discovered rather than planned for, so it is worth stating before a first deployment rather than after.

  • Unbounded runs. Without a stopping condition and spend limits, an agent that cannot complete its goal keeps trying. The result is cost and latency rather than an error, so nothing alerts.
  • Correct reasoning on a wrong premise. An agent that reads an ambiguous field, interprets it plausibly, and acts is not malfunctioning. It reached a wrong conclusion competently, which is invisible to model testing. See data intelligence.
  • Inherited permissions. Agents frequently run as the identity that created them, which is usually far broader than the task needs. The over-permissioning is silent until something reaches further than intended.
  • No named owner. When an agent misbehaves, the first question is who is responsible for it, and in many estates nobody can answer. See agent registry.
  • No usable record. Answering what an agent did, on what data, under whose authority requires per-action logging captured at the time. It cannot be reconstructed later, which becomes a compliance problem the first time it is asked for. See AI governance.

The pattern across all five is that they are operational rather than technical. An agent that reasons well and is deployed without limits, permissions, ownership, and a record is a well-built component inside a badly built system. That gap between the two is what the rest of this glossary is about, and the practice of closing it is AgentOps.

Frequently asked questions about AI agents

If your question is not here, our team will answer it directly.


Talk to a Specialist →
What is an AI agent in simple terms?

Think about the difference between a form and an assistant. A form collects what you give it, in a fixed order, and cannot cope with anything unexpected. An assistant works out what to do next based on what it finds. An AI agent is software that behaves like the second: you give it an objective rather than instructions, and it decides which steps to take to reach it, using systems and data along the way. Conventional software follows a path. An agent chooses one.

What is the difference between an AI agent and agentic AI?

An AI agent is a thing; agentic AI is a description. The agent is the individual software system that pursues a goal and acts. Agentic AI is the broader category, covering the approach, the architecture, and the class of systems built from one or more agents. In everyday use, people say AI agent when discussing a specific system and agentic AI when discussing the shift in how software is built. Neither is a subset of the other in any formal sense, because one is a noun and one is a category.

How do you tell whether something is really an AI agent?

Ask who decided the order of the steps. If a developer or business analyst wrote the sequence in advance, it is automation, however much AI sits inside any individual step. If the software works out the sequence while running, it is an agent. Three properties usually travel with that answer: it can take actions rather than only produce output, it attempts cases nobody enumerated, and it decides when it is finished. The third is the one most often missing from things sold as agents.

What are the components of an AI agent?

Five: a goal, a reasoning model, memory, tools, and a stopping condition. The goal is an objective rather than a list of steps. The reasoning model, usually a large language model, decides what to do next. Memory holds what has happened, within a run and sometimes across runs. Tools are the systems it can actually reach and act on. The stopping condition defines done, plus limits on steps, time, and spend. The fifth is routinely omitted from vendor diagrams and is behind most runaway cost incidents.

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

A chatbot converses; an agent acts. A chatbot answers questions turn by turn, with the user driving each exchange, and its job finishes when it has replied. An agent is given an objective and works toward it across several steps, calling systems and changing things, often without a person present for each decision. Some products blur the line by adding a small number of actions to a conversational interface, which is why the useful test is how much the software decides rather than what its interface looks like.

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

A copilot keeps a person in control; an agent does not, by design. A copilot suggests, drafts, and completes work inside somebody's workflow, and every output passes through a human who accepts or rejects it. That review is itself a control. An agent takes the action directly, which removes that control and means something else has to replace it: permissions, guardrails, spend limits, and an audit record. The difference is less about capability than about where accountability sits.

What are the main types of AI agents?

The classical taxonomy from AI research grades them by how they decide: simple reflex agents react to current input with no memory; model-based agents maintain an internal picture of the world; goal-based agents plan toward a stated objective, which is where most enterprise agents sit; utility-based agents weigh options against a measure of value; and learning agents adjust based on outcomes. Enterprise conversations more often group agents by function instead, such as service, procurement, or reconciliation agents, which is better for scoping and worse for architecture.

Do AI agents replace RPA?

Not generally, and the two suit different problems. RPA repeats a recorded sequence exactly, which makes it predictable, cheap to run, and brittle when an interface changes. An agent works out its own path, which makes it resilient to variation and correspondingly harder to predict. For a high-volume process with a stable shape, RPA remains the better engineering choice. Agents earn their place where the process has genuine variation, where exceptions are frequent, or where judgement is needed at some step.

Are AI agents safe to use in production?

They can be, and the safety comes from how they are deployed rather than from the model. Five failures recur in practice: runs with no stopping condition, correct reasoning over a wrong premise, permissions inherited from whoever created the agent, no named owner, and no per-action record. Every one is an operational gap rather than a reasoning failure. An agent with bounded scope, narrow permissions, a spend limit, a named owner, and a complete audit trail is a manageable system. The same agent without those is not.

What can AI agents actually do today?

They work well on multi-step processes that cross systems, involve frequent exceptions, and have a checkable outcome. Common production examples include triaging security alerts, reconciling records across finance systems, handling support cases end to end, screening supplier documents, and researching then drafting a document. They work considerably less well where the outcome cannot be verified, where an error is irreversible and unreviewable, or where the underlying data is ambiguous, because in that last case the agent reasons correctly and still reaches a wrong answer.

How much autonomy should an AI agent have?

Match it to the reversibility of the action rather than to the sophistication of the technology. Reading, summarizing, and drafting are low-risk and can run unattended. Actions that move money, change a customer record, or communicate externally warrant a checkpoint until the agent has a track record. The practical approach is to start with a human confirming actions, measure how often the confirmation changes the outcome, and remove the checkpoint only where that rate is genuinely near zero.

Do you need one agent or several?

Start with one and add more only when a single agent is failing for a specific reason. Several specialized agents help where a process needs genuinely different expertise at different steps, or where parts can run in parallel. They cost you reliability, because component accuracy multiplies: five agents at ninety-five percent each is well under eighty percent end to end. Vendors tend to present multi-agent designs as more advanced. They are more capable and less reliable, and the trade is worth making deliberately rather than by default.

From one agent to an estate
Your first agent is an experiment. Your fiftieth is an estate.

CAMS covers what changes between the two: a registry with a named owner per agent, promotion through Dev, QA and Production, guardrails and spend limits enforced on every action, and an audit record you can query rather than reconstruct.