Lumo Copilot

An AI copilot that runs entirely on your Mac. Your files, your model, your machine — nothing leaves your computer.

No account. No cloud. No subscription.

Apple Silicon · macOS 11+ · free and open

Private by construction, not by promise

Most AI assistants send your conversations — and increasingly your files — to someone else's servers. Lumo makes the opposite bet: the model runs on your machine, so there's nothing to send.

Conversations and files never leave your Mac
No account, no sign-in, no telemetry
Works with no internet connection at all
Every file change and command asks you first
The local API is loopback-only and origin-guarded
Open source — read exactly what it does

What it does

A real assistant, not a chat box — it can read your files and act on your Mac, but only ever with your approval.

💬

Private chat

Talk to an open model — Gemma, Llama, Granite, Phi and more — running locally. Attach a document and it reads it directly, no upload.

Acts on your Mac

Organise folders, dig into disk usage, run commands. It shows you exactly what it will do and waits for your OK every time.

🎛️

Knows its own limits

Models are sorted by what they can actually do, and tested by hand. Lumo never claims an ability the model you picked doesn't have.

🔍

Deep Research

Ask a real question. Lumo plans searches, reads the pages it finds, and writes one answer with citations back to every source.

⚖️

Compare

Ask several of your models the same thing and read the answers side by side. Find out which one is actually good at your work.

⌨️

Coding workspace

File tree, editor, terminal, git diff and a live preview — with the assistant beside it, able to read and change the project.

🧩

Extensible (MCP)

Add tool servers over the open Model Context Protocol. You choose each one, and every call still asks permission.

📚

Projects & memory

Group related chats with shared context, notes and reference files, so the assistant keeps its bearings across sessions.

Schedules

Hand it recurring chores. It can even wake the Mac to run them, if you let it.

A look inside

The Lumo Copilot chat window running a local model
Chat runs on a model installed on your own machine.
The model picker showing capability categories and memory fit
Pick a model by what it can do — and see whether it fits your Mac's memory.

Getting started

About five minutes, most of it waiting for a model to download.

  1. Download the DMG and drag Lumo to your Applications folder.
  2. Approve it once. Double-click Lumo — macOS will refuse it the first time. Open System Settings → Privacy & Security, scroll to Security, and click Open Anyway. You only do this once.
  3. Let it set itself up. On first run Lumo offers to install Ollama, which runs the models.
  4. Pick a model. Lumo recommends ones that suit your Mac's memory and tells you honestly what each is good at. Gemma 4 · 12B is a solid start.
  5. Say hello. Then try something real: what's eating my disk space?

Why the prompt? Removing it requires a paid Apple Developer certificate. Lumo is free, so macOS shows its standard warning for apps from outside the App Store. Prefer the terminal? xattr -dr com.apple.quarantine "/Applications/Lumo Copilot.app" does the same thing.

Ready to try it?

Free, open, and yours. It updates itself when there's something new.

Apple Silicon (M1 and later) · macOS 11 Big Sur or newer · about 8 GB free for a model