You can run an AI chatbot on your own computer. Here's how.

Plenty of people would like to use an AI assistant but don't want their questions, documents or family details sitting on someone else's servers.

It's a fair worry, and these days you can do something about it. There are free programs that download an AI model onto your own computer and run it there. Once it's downloaded it works with the internet unplugged.

What you give up

A model that fits on a home computer is smaller than the ones behind the big online services. For everyday writing jobs, like drafting a letter or summarising a page you paste in, a small model is often good enough. It's weaker at hard reasoning, and since it isn't looking anything up online, it knows nothing that happened after it was made. Try it on a small job before you rely on it.

What you need

Check memory first: RAM on a Mac, and RAM plus the graphics card's memory on a PC. The whole model has to fit in it while it runs.

LM Studio, one of the two apps below, publishes these recommendations:

• Mac: an Apple silicon Mac (M1 or newer) running macOS 14 or later. 16 GB of memory is recommended. 8 GB can work with smaller models.
• Windows: 16 GB of RAM recommended, and at least 4 GB of dedicated graphics memory.
• Intel Macs aren't supported by LM Studio.

Models also take up disk space. Google's Gemma 4 is a good place to start, and Ollama's library lists its smaller versions at roughly 5 to 10 GB each to download. As a rough rule, if the download is bigger than your free memory, pick a smaller version.

Option 1: LM Studio (if you'd rather click than type)

LM Studio is a regular app with a chat window, a lot like the online services. In 2025 the company announced it's free to use at home and at work.

1. Download it from lmstudio.ai and install it like any other app.
2. Open the Discover tab and search for a model by name, such as Gemma 4. Pick a size that fits the memory rule above.
3. Download one, load it, and start typing.

Option 2: Ollama (if you don't mind a terminal)

Ollama is free and works on Mac, Windows and Linux. You download the installer from ollama.com, and after that you mostly use it by typing commands.

To download Gemma 4 and start chatting, open Terminal on a Mac (or PowerShell on Windows) and type:

ollama run gemma4

The first run downloads the model. After that it starts in seconds. Type your question, press Return, and type /bye when you're done.

Two more commands are worth knowing:

• ollama list shows the models you've downloaded.
• ollama ps shows what's running. The Processor column says whether the model is on your graphics chip ("100% GPU", fast) or has spilled into ordinary memory ("CPU", much slower). If you see CPU, try a smaller model.

Where it all lives

The models are just big files on your disk. Ollama keeps them in a folder called .ollama in your home folder on a Mac or Windows PC. If you run short of space later, deleting models you no longer use gets it back. ollama rm followed by the model name does it.

What I do about it

If you'd like to try this but don't want to do the setup yourself, I can install it, pick a model that suits your computer, and show you how to use it in one visit. I'll also tell you plainly if your computer isn't up to it, rather than leave you with something that crawls.

Call (503) 567-5975 or email support@ironwood.help.

Sources

• LM Studio: System requirements
• LM Studio: LM Studio is free for use at work (July 8, 2025)
• Ollama: Download
• Ollama: Gemma 4 model page, with model sizes and the ollama run commands
• Ollama: FAQ, covering local privacy, where models are stored, and reading ollama ps
• Ollama: Cloud models, which run on Ollama's servers and require sign-in
• Ollama: CLI reference and the interactive chat source, which documents /bye to exit