Guide · Open source

How to use Omi: install the macOS app, Windows setup and tips

Updated 2026-10-29 · written with AI from the project's own documentation and checked against our catalogue data

Omi is an open-source app from BasedHardware that captures your screen and conversations, transcribes them in real time, and produces summaries and action items. It also has an AI chat that can answer questions about what you have seen and heard. This guide shows how to get the project running from source on macOS or Windows, and what to know before you do.

omiBasedHardware/omi · ★ 13.9k · MIT Install options & downloads

Quick start

1Check your system. On macOS, confirm you have macOS 14 or later, Xcode and Node.js. On Windows, confirm Node.js is installed.
2Clone the repository. Run git clone on the BasedHardware/omi repository from GitHub.
3Launch the macOS app. Move into omi/desktop/macos and run ./run.sh --yolo to build and start the app.
4Or set up the Windows app. Move into omi\desktop\windows, run npm install, copy .env.example to .env and run npm run dev.
5Review the privacy settings. Decide whether you are comfortable with continuous screen and audio capture before you rely on it for real conversations.

What Omi does

Omi captures your screen and conversations, transcribes them in real time, generates summaries and action items, and provides an AI chat that can answer questions about what you have seen and heard.

It runs on desktop, phone and an optional wearable device. The project is published under the MIT license and is written mainly in Python.

Install Omi on macOS

The quickest route is the macOS quick start in the README. It clones the repository, moves into the macOS desktop folder and runs the launch script. According to the README, this builds the macOS app, connects to the cloud backend and launches, with no env files, credentials or local backend needed.

Requirements: macOS 14 or later
Xcode, which includes Swift and code signing
Node.js

git clone https://github.com/BasedHardware/omi.git && cd omi/desktop/macos && ./run.sh --yolo

Run Omi on Windows from source

The Windows desktop app is started from source. Clone the repository, move into the Windows folder, install dependencies, copy the example environment file and start the dev app. The README says this uses the public config in .env.example.

The only listed requirement for Windows is Node.js.

git clone https://github.com/BasedHardware/omi.git
cd omi\desktop\windows
npm install
copy .env.example .env
npm run dev

Set up a development worktree or the full backend

For development worktrees, the README recommends running a baseline local setup once. It installs the Git hooks and syncs the pinned backend Python environment used by selected pre-push checks. Mobile and desktop runtime environments are opt-in.

The full installation is for local development with the complete backend stack. It needs the Xcode command line tools and uv, and it copies the backend example environment file. The README points to desktop/macos/README.md for environment variables and credential setup.

Run make setup once for the baseline setup
Use python3 scripts/worktree_env_doctor.py --all --fix to prepare app, backend and desktop state in a new worktree
Mobile app setup: cd app and run bash setup.sh ios or bash setup.sh android

make setup

Use the mobile app and the wearable

The README lists mobile apps for iOS and Android, and a browser option at app.omi.me. The optional Omi wearable pairs with the mobile app for continuous capture. The README describes it as an open-source AI wearable that supports 24h+ continuous capture.

Hardware pages in the README cover buying, building, flashing firmware and integrating a wearable. These steps are for developers and need hardware setup.

Browser: app.omi.me
Mobile: App Store and Google Play links are in the README
Wearable: buying guide, build instructions and firmware flashing docs are linked from the README

Build your own apps and integrations

The README links to a developer documentation set covering app development, audio streaming apps, custom chat tools and an API reference for memories, conversations and action items. There is also a device SDK set for Python, Swift and React Native, and an MCP Server for Model Context Protocol integration.

If you plan to modify the code, start with the architecture docs listed in the README: backend deep dive, transcription pipeline, chat system and BLE protocol.

API reference: REST endpoints for memories, conversations and action items
Python device SDK with BLE, Opus and Deepgram support
MCP Server in the mcp/ folder

What people use it for

Keeping a searchable record of meetings and conversations
Getting automatic summaries and action items after calls
Asking the AI chat what you saw or heard earlier in the day
Developing or modifying a wearable AI product using the SDKs and docs
Building custom apps or chat tools through the developer documentation

Frequently asked questions

Is Omi free and open source?

The README says Omi is fully open source and the project is published under the MIT license. The license text is in the LICENSE file in the repository. The README does not list prices, so check the project's website for any paid products or hardware.

What are the requirements to install Omi?

For the macOS quick start, the README lists macOS 14 or later, Xcode and Node.js. For the Windows setup, it lists Node.js. The full backend setup also uses uv.

Does Omi work offline or record locally?

No. The README says the macOS quick start connects to the cloud backend. Omi is not a fully local, offline recorder, so it is not a fit if you need that.

Is Omi safe for privacy?

Omi captures your screen and audio continuously, as the README describes. If that is uncomfortable for you, or for people in your conversations, think carefully before using it. The README does not give detailed privacy guarantees, so read the project's documentation.

Why does the Windows install need developer tools?

The Windows setup runs from source. It requires Node.js, npm install and npm run dev, so it is not a one-click installer.

What are alternatives to Omi?

The project's analysis compares it with Rewind, Screenpipe, Limitless and Otter.ai. Those tools overlap with Omi in recording, transcription or meeting notes, but this page does not compare their features or prices.

New agent skills, repos and AI tools — every week

The best new skills and open-source projects, the fastest-growing repos, and one free prompt. Five minutes to read. Unsubscribe anytime.

Our verdict

Omi suits people who want a searchable record of meetings and conversations, and developers who want to build or modify a wearable AI product. Skip it if you need fully local, offline recording, if you are uneasy about constant screen and audio capture, or if you want a simple one-click install without developer tools.

Install options & downloads