Give It
Tools

Beyond Text-In, Text-Out

Think about what happens when you ask a chatbot: "What time is it in Tokyo right now?" The AI doesn't actually know. It was trained on data that's months or years old. It has no clock. So it does the only thing it can — it guesses, based on patterns in its training data. Sometimes it's close. Sometimes it's hilariously wrong.

Now give that same AI a tool — a simple function that returns the current time in any timezone. Suddenly:

Same question. Completely different reliability. The AI didn't get smarter — it got connected. It can now reach outside its own brain and interact with the real world.

How Tool Use Actually Works

Here's the mental model. When you give an AI access to tools, you're handing it a toolbox and saying: "Here are things you can do. For each tool, here's what it's called, what it accepts, and what it returns. You decide when to use them."

The AI doesn't blindly use every tool on every request. It reasons about which tools are relevant, decides whether to use one, formulates the right input, calls the tool, reads the result, and then incorporates that result into its response. This process is called function calling — and it follows a remarkably consistent loop.

This is the agent loop — observe, think, act, observe, respond. And it's the same loop whether the AI has one tool or a hundred. The more tools you give it, the more capable it becomes. But also, the more trust you need to place in its judgment about which tool to use and when.

Seven Tools That Change Everything

There are hundreds of specific tools AI can use, but they mostly fall into seven categories. Each one extends what AI can do in a fundamentally different way. Some give it knowledge (web search, databases). Some give it abilities (code execution, image generation). Some give it hands (file editing, API calls).

Click any tool below to see the AI's full reasoning process — from recognizing it needs a tool, to calling it, to using the result.

The Trust Question

Once AI can do things — not just talk about them — a new question emerges. A question that has nothing to do with technology and everything to do with you:

This isn't a technical question — it's a human one. And the answer depends on three things:

The Tool-Use Mindset

There's a subtle shift that happens when you start thinking about AI as a tool-user instead of just a text-generator. You stop asking "What can the AI say about this?" and start asking "What can the AI do about this?"

Instead of asking the AI to describe how to analyze data, you give it a code execution tool and watch it write and run the analysis. Instead of asking for links to research, you give it web search and let it find and synthesize the information itself. The output goes from "here's what you could do" to "here's what I just did."

The first prompt gets you a tutorial. The second gets you an answer. With tools, the AI goes from advisor to assistant — from someone who tells you what to do, to someone who actually does it.

Key Concepts

In the next chapter, we take this further. Tools are the building blocks. Now it's time to wire them together into something that can pursue goals on its own. It's time to build agents.