Install and Chat with Ollama

Time to actually run AI on your machine. In this lesson you will install Ollama, a free and open-source tool that downloads models and runs them for you, and you will have a real conversation with an AI model that lives entirely on your laptop. Ollama is one of the most popular ways to run local AI, and it is friendly to beginners: it now includes a simple chat window, so you can use it without touching the terminal if you prefer.

We will cover both the point-and-click way and the one-line terminal way, so you can use whichever feels comfortable.

What You'll Learn

• How to download and install Ollama on Windows or Mac
• How to chat using the built-in desktop window (no terminal needed)
• The one terminal command that downloads and runs a model
• How to manage the models you have downloaded

Step 1: Download and Install

Go to ollama.com and click the download button. It will detect your system and offer the right version for Windows, Mac, or Linux. On Mac, Ollama requires a recent macOS version, so update your system if it prompts you.

The install is a normal app install: open the downloaded file and follow the prompts. When it finishes, Ollama starts quietly in the background and adds a small icon to your menu bar (Mac, top-right) or system tray (Windows, bottom-right near the clock). That background piece is the engine that actually runs models. You do not need to configure it.

Step 2: The Easy Way, the Desktop Chat Window

Recent versions of Ollama include a built-in chat window that looks and feels like ChatGPT, except it runs locally. This is the friendliest way to start.

1. Click the Ollama icon in your menu bar or system tray, or open the Ollama app from your applications.
2. A chat window opens. At the top there is a model dropdown.
3. Pick a model from the dropdown. If it is not downloaded yet, Ollama fetches it the first time (a one-time wait of a few minutes).
4. Type a message at the bottom and press Enter. The reply is generated on your own computer.

That is it. You are chatting with a private, local AI. Try asking it to "write a short thank-you email to a coworker" or "explain how a bicycle gear works in simple terms." If a small model feels limited, remember you can switch models from the same dropdown.

Step 3: The One-Line Terminal Way

Ollama also has a famously simple command-line method, and it is worth knowing because a lot of guides use it. Do not be intimidated; it is a single line.

Open your terminal:

• Windows: press the Windows key, type "PowerShell," and open it.
• Mac: press Command and Spacebar, type "Terminal," and open it.

Now type this one command and press Enter:

ollama run llama3.1

Here is what happens. The first time, Ollama downloads the model (you will see a progress bar). Once it is ready, you get a prompt where you can type your message and get a reply right there in the terminal window. To leave the chat, type /bye and press Enter.

llama3.1 is just one example of a model name. You could swap in another model, for example ollama run gemma3 or ollama run phi4. Browse the full catalog at ollama.com/library, where models are listed with their sizes so you can pick one that fits your memory from the previous lesson.

Step 4: Manage Your Models

A few more simple commands help you stay tidy. Type these in the same terminal:

• ollama list shows every model you have downloaded and how much space it uses.
• ollama pull <model-name> downloads a model without starting a chat, handy for getting one ready ahead of time.
• ollama rm <model-name> deletes a model you no longer want, freeing up disk space.

Everything you download through the terminal also shows up in the desktop chat window's dropdown, and the other way around. They share the same engine, so you can mix and match freely.

A Note on Privacy

Once a model is downloaded, all of this runs on your machine. You can turn off your WiFi and the chat still works. The only time Ollama needs the internet is to download a new model or update itself. Your conversations are never uploaded.

If Something Goes Wrong

• The command is not recognized: close the terminal and open a new one so it picks up the freshly installed tool. If it still fails, restart your computer once after installing.
• The download is slow or stalls: it depends on your internet speed and the model size. Let it finish, or try a smaller model first.
• Replies are very slow: you likely picked a model too big for your memory. Try a smaller one (for example a 3B model). The next lesson covers sizing in detail.

Key Takeaways

• Install Ollama from ollama.com; it runs quietly in the background as the engine for local models.
• The easiest way to chat is the built-in desktop window: pick a model from the dropdown and type, no terminal required.
• The classic terminal method is one line: ollama run <model-name>, then chat, then /bye to exit.
• Manage models with ollama list, ollama pull, and ollama rm. Browse models at ollama.com/library.
• After the one-time download, everything runs offline and privately on your machine.