How to run a private AI assistant on your Mac with Ollama

What you will end up with

By the end of this guide you will have OpenDot running on your Mac with a local model served by Ollama. Your Dots will think on your own machine, with no API key, no per-token cost and nothing sent to a cloud model.

OpenDot is in early access, so there is no signed download yet. This guide builds it from source, which takes a few minutes and is the same route the developers use.

What you need

• A Mac running macOS 14 Sonoma or later, Apple silicon or Intel.
• Node.js 22.19 or later (from nodejs.org, or brew install node).
• Git, which comes with the Xcode Command Line Tools (xcode-select --install).
• Ollama, from ollama.com.

Steps

1. 1

   Install Ollama and pull a model

   Install Ollama from ollama.com, then download a model. llama3.1 is a good first choice, and you can use any model Ollama offers.

   ollama pull llama3.1

2. 2

   Get the OpenDot code

   Clone the repository and install the app's dependencies.

   git clone https://github.com/athulsreekumar/opendot.git
   cd opendot/desktop
   npm ci

3. 3

   Run OpenDot

   Start the app. A short setup opens the first time.

   npm run dev

4. 4

   Pick Ollama as your model

   During setup, choose the free and private option. If Ollama is running with a model, OpenDot finds it automatically. You can check it later in Settings, Models, where Ollama appears under On this Mac. Press Test to confirm it responds and Discover models to list what you have pulled.

5. 5

   Create your first Dot

   Describe the job in one sentence, for example "Watch my inbox and tell me what actually needs me", tick the tools it may use and let OpenDot write its name and personality.

6. 6

   Optional: choose a model per Dot

   The default model applies to every Dot, but you can give any Dot a different one, so a quick local model can handle routine jobs while a larger one handles research.

Turn it into a real Mac app

npm run dev is perfect for trying things out. When you want an app in your Applications folder, build a DMG for your chip:

npm run dist:mac:arm64   # Apple silicon
npm run dist:mac:x64     # Intel

The first-launch warning

Early-access builds are not notarized by Apple yet, so macOS warns you the first time. Right-click OpenDot in Applications, choose Open, then Open again. Or run xattr -dr com.apple.quarantine /Applications/OpenDot.app once.

What stays on your Mac, and what does not

With Ollama, the model runs on your machine and the Dot's thinking never leaves it. Your Dots' settings, memory and chats live in ~/.opendot as plain files. Two honest limits: Dots that read Gmail, Outlook or other online services still need an internet connection to reach them, and Dots only run while OpenDot is open and your Mac is awake.