What Is Agentic AI?

Audience

ages 10–15

Time

20 to 30 min

Module

1/5

Badge to earn

Agent Spotter

Today's mission

By the end of this lesson, you will be able to explain the difference between a chatbot and an agent, and name the three extra ingredients that turn a talking AI into an AI that acts.

The idea in one picture

Imagine you ask a friend, "What's the weather like?" They answer, "Sunny, 22 degrees," and the conversation is over. That's a chatbot: one question, one answer, done. Now imagine you say to an assistant, "Plan my trip to Paris." They don't just talk: they check flights, compare hotels, book the best one, and come back to tell you, "All set, here are your tickets." They took real steps, one after another, until your goal was reached. That's an agent. A chatbot answers; an agent gets things done.

How it works

A regular chatbot answers one question and stops. An agent keeps going: it thinks, takes actions, checks what happened, and keeps working until the job is actually done.

Read the diagram like this: on the left, the chatbot's path is a single arrow. On the right, the agent's path passes through many steps before arriving at a result. The agent is built from the same kind of AI model as the chatbot, but three extra ingredients are added around it:

• Tools — it can search the web, run code, send an email, read a file. The model can't do these things alone, so it "picks up" a tool.
• Memory — it remembers what it already tried, so it doesn't repeat the same mistake or redo work it already finished.
• A loop — it keeps working until the goal is met, instead of stopping after one reply. Each time around the loop, it checks: "Am I done yet?"

Examples:

• A chatbot: you ask, "What's the capital of France?" It answers, "Paris." Finished.
• An agent: you say, "Book me the cheapest flight to Paris next month." It searches several websites, compares prices, picks the cheapest one that fits your dates, and actually books it. Then it tells you what it did.

Notice the second example has a goal, not a question. Goals need many steps, and steps need tools, memory and a loop. That is exactly what makes something "agentic."

Screen-free activity

Chatbot or agent? Grab a pencil and a friend or family member.

1. Write down 6 requests on small slips of paper, for example: "What time is it?", "Organize my birthday party", "How do you spell giraffe?", "Find me a good book about dragons and get it from the library", "Is 7 a prime number?", "Clean my room".
2. Shuffle the slips. Take turns picking one and deciding out loud: would a chatbot be enough (one answer) or do you need an agent (many steps)?
3. For each "agent" slip, list at least 3 steps the agent would have to take and one tool it would need.
4. Discuss: is there a request where you would rather do the steps yourself than let an agent do them? Why?

Key takeaways

• A chatbot gives one answer to one question; an agent works toward a goal through many steps.
• An agent adds three ingredients to the AI model: tools, memory and a loop.
• Agents act in the real world, so what they do (booking, sending, buying) actually happens.

A question for a grown-up

If you could hand one boring task in your daily life to an agent, which one would it be, and what would you want to double-check before letting it act?

So what? Agentic AI is not just a smarter chatbot. It is an AI that can act in the world, step by step, until your goal is achieved — which is exciting, and also exactly why you'll learn how to keep it safe in the rest of this course.