explaingit

basedhardware/omi

★ 12,517DartAudience · generalComplexity · 4/5LicenseSetup · moderate

TLDR

Open-source AI wearable and app that records your conversations all day, transcribes them in real time, and gives you an AI that remembers everything, works as a mobile app, desktop app, or paired with an optional Bluetooth clip-on device.

Mindmap

mindmap
  root((omi))
    What it does
      Continuous recording
      Real-time transcription
      AI memory chat
      Action item summaries
    Form factors
      iOS and Android app
      macOS desktop app
      Bluetooth wearable
      Omi Glass camera
    Tech stack
      Python FastAPI
      Firebase storage
      Deepgram audio
      Dart mobile
    For developers
      REST API
      Python and Swift SDKs
      MCP server
Click or tap to explore — scroll the page freely

Code map

Detail Auto

An interactive map of this repo's files and how they connect — its source is parsed live in your browser. Click Visualize to build it.

filefunction / class

Things people build with this

USE CASE 1

Record and automatically summarize every meeting and conversation throughout your day without taking manual notes.

USE CASE 2

Build a third-party integration using the REST API to connect your Omi memories to Notion, Slack, or other productivity tools.

USE CASE 3

Run the open-source mobile app as a free personal AI memory system without purchasing the hardware device.

USE CASE 4

Connect Omi conversation memories to an AI coding assistant via the Model Context Protocol server.

Tech stack

PythonFastAPIDartFirebaseDeepgramESP32

Getting it running

Difficulty · moderate Time to first run · 30min

Mobile app works without hardware, the physical Omi Bluetooth device or ESP32 Omi Glass accessory is optional for full wearable capture.

MIT license, use freely for any purpose including commercial projects, as long as you keep the copyright notice.

In plain English

Omi is an open-source AI system that captures what you see and hear throughout the day, then lets you chat with an AI that remembers all of it. It is described as a second brain: you run it in the background, and it transcribes conversations in real time, generates summaries and action items, and stores everything in a searchable memory that persists across sessions. The system works across multiple form factors. A macOS desktop app captures screen activity and audio from your computer. A mobile app for iOS and Android pairs with optional wearable hardware. The wearable itself is a small Bluetooth device sold by BasedHardware that clips on or hangs around the neck and records ambient audio continuously for over 24 hours on a battery charge. A companion device called Omi Glass is an ESP32-based camera and microphone accessory for those who want to capture video as well. All the hardware designs are open source. The backend is written in Python using FastAPI. Audio streams are transcribed using Deepgram, processed for voice activity detection and speaker identification, and then summarized and analyzed by large language models. Conversation history and memories are stored in Firebase. The entire pipeline runs in real time. Third-party developers can build integrations using a REST API, SDKs for Python, Swift, and React Native, and a Model Context Protocol server for connecting Omi memories to AI coding assistants and agents. The project is released under the MIT license and reports over 300,000 users. It can be tried in a browser without hardware at app.omi.me.

Copy-paste prompts

Prompt 1
Using the Omi REST API, show me how to retrieve today's conversation summaries and action items from my account in Python.
Prompt 2
Help me build a Python integration that reads new Omi conversation summaries and automatically creates Notion pages from them.
Prompt 3
Using the Omi SDK, how do I connect my Omi memories to a Claude-based AI assistant via the Model Context Protocol server?
Prompt 4
Show me how to set up the open-source Omi mobile app on iOS without the hardware device and start recording conversations.
Open on GitHub → Explain another repo

← basedhardware on gitmyhub — every repo by this author, as a profile.

Verify against the repo before relying on details.