What Is an AI Agent?
- Define an AI agent in plain terms without vendor framing
- Explain the key difference between an agent and a chatbot
- Identify the three additions that turn an LLM into an agent
The Word Gets Used for Everything
Open any tech news site and you'll find "AI agent" attached to chatbots, automated scripts, and entire software platforms. The term has been stretched so thin that almost anything with a prompt inside it now claims the label. That makes it harder, not easier, to understand what agents actually are.
This lesson cuts through the noise with a definition grounded in what agents actually do — not what vendors claim.
The Core Definition
An AI agent is a software system that perceives its environment, makes decisions, and takes actions to achieve a goal — with limited human input at every step.
That last part matters. A system that waits for a human to approve every move is not autonomous. An agent operates within defined boundaries but chooses its own path inside them.
Cognizant's AI research describes the shift this way: a generative AI system becomes an agent once it moves past producing output from a prompt and begins acting on information to reach an outcome.
What Makes It Different From a Chatbot
A chatbot responds. An agent decides what to do next.
When you ask a chatbot a question, it generates an answer and stops. When you give an agent a goal, it figures out the steps, calls whatever tools it needs, checks whether the result is good enough, and keeps going until the job is done — or until it hits a limit you set.
- Chatbot — receives a message, produces a reply, stops
- AI agent — receives a goal, plans steps, calls tools, checks results, continues until done
What Makes It Different From an LLM
An LLM (large language model) is the "brain" that generates text. On its own, it's a prediction engine: given input, it outputs the most likely continuation. That's powerful, but it's not agentic.
An AI agent wraps that brain with:
- Memory — so it can remember context across steps
- Tools — so it can search the web, run code, write files, call APIs
- A loop — so it can keep working until the goal is reached
Remove any one of those three and you have a very capable text generator, not an agent.
A Minimal Working Definition
For this course, an AI agent is any system where a language model autonomously decides the next step, can call tools to execute that step, and repeats the cycle until a goal is complete.
That definition is tight enough to exclude plain chatbots and LLM wrappers, but broad enough to cover the wide range of agents you'll encounter — from simple single-step automation to complex multi-agent pipelines.
Why the Term Gets Abused
Agents are a marketing category as much as a technical one. Calling a product an "AI agent" implies autonomy, intelligence, and business value — so vendors apply it liberally. A good test: does the system make decisions about what to do next, or does it just execute what a human told it to do? If it's the latter, it's not really an agent regardless of the label.
- An AI agent is a system that perceives its environment, makes decisions, and takes actions toward a goal with limited human input at each step
- The defining property is autonomy: the agent chooses what to do next — not a human, and not a fixed script
- LLM + memory + tools + a loop = agent; remove any one of those and the system is not truly agentic
- The term is heavily overloaded — the real test is whether the system makes decisions about what to do next, not just executes what it was told
Progress is saved in this browser only — clear your cache or switch device and it's gone. Create a free account and it follows you everywhere.