MoMo Agent

Make Life Your Context — An Open-Source, Self-Hosted 24/7 Personal Agent for Apple Watch, iPhone & Mac

Free On-Device Speech Transcription · Powered by Your Codex, Claude, or Antigravity Subscription · Self-Organizing Living Memory

English | 简体中文

MoMo Agent is an open-source, fully self-hosted all-day personal agent built natively for Apple Watch, iPhone, and Mac.

Capture your day effortlessly from your Apple Watch or Bluetooth earphones, transcribe speech 100% locally on your device at $0 cost, and let Momo on your Mac turn fleeting conversations into self-organizing long-term memory, timely reminders & morning briefings, and illustrated daily life journals—using the Codex, Claude, or Antigravity subscription you already have, without ever sending your life data to a third-party cloud.

_{Your devices, working together. Models shown are illustrative. Model credits & noncommercial artwork license}

Why MoMo Agent?

• A Living Memory That Organizes Itself: Stop filing notes by hand. MoMo quietly distills your daily conversations into an Obsidian-style multi-level knowledge vault—connecting people, projects, and ideas automatically, keeping every memory one tap away from the original voice moment, and updating both notes and speaker names when you ask in plain language.
• A Proactive Butler That Only Acts for You: Wake up to a concise Morning Briefing, and let Momo turn your spoken commitments into Apple Reminders, Calendar events, and Notes—anchored to your voiceprint so another speaker's remarks never end up on your to-do list.
• Your Day as an Illustrated Visual Journal: Turns meaningful moments and real places you visited into warm kiss-cut sticker illustrations and an interactive daily footprint map—making your life log something you genuinely enjoy looking back on, not a wall of raw transcripts.
• $0 Extra AI Bills — Reuse the Subscription You Already Own: An all-day personal agent shouldn't rack up surprise API bills. Sign in with OpenAI Codex, Claude, or Google Antigravity (or any custom API), switch providers anytime in one click, and transcribe unlimited audio locally on-device for $0.
• 100% Self-Hosted, Private & Effortless Capture: Record naturally from your wrist on Apple Watch (with fast local Wi-Fi sync) or via Bluetooth earphones, and stay privately connected to your Mac from anywhere—your audio, voiceprints, and plain-Markdown memories never leave your own devices.

Product Preview

1. Wrist Capture, Proactive Butler, Illustrated Journal & Mobile Memory (watchOS + iOS)

• 1. Apple Watch All-Day Capture: Hands-free wrist recording with live waveform, one-tap mute & bookmark, and fast automatic sync to iPhone.
• 2. Proactive Butler & Morning Briefing: Start each day with a tailored briefing and ask Momo anything across your conversations, tasks, and schedule.
• 3. Illustrated Journal & Timeline: Browse daily life illustrations, footprint map, and speaker-aware transcripts with silence-skipping playback.
• 4. Self-Organizing Memory Vault: Explore multi-level project, people, and journal notes on the go, jump back to original audio, or refine memories in one sentence.

2. Self-Hosted Mac Hub, Living Memory Reader & Flexible AI Subscriptions (macOS)

• Left — Proactive Steward & Closed-Loop Actions: Hear your spoken commitments turn into Apple Reminders automatically, or tell Momo in one sentence to split a note and sync a speaker voiceprint back to iPhone.
• Right — Multi-Level Memory Vault: Browse your self-organized Markdown knowledge tree, follow bidirectional links and backlinks, replay cited voice clips, and switch across Codex, Claude, or Antigravity subscriptions anytime.

How Your Devices Work Together

flowchart LR
    W["Apple Watch / Earphones\nAll-Day Hands-Free Capture"] --> P["iPhone\nFree On-Device Transcription\nJournal · Footprint Map · Mobile Memory"]
    P <-->|"Encrypted Private Mesh\n(Built-in Tailscale / Local Wi-Fi)"| M["Mac (Self-Hosted Hub)\nProactive Butler · Living Markdown Vault\nReminders / Calendar / Notes · Life Illustrations"]
    M <--> AI["Your Existing Subscription\nCodex · Claude · Antigravity"]

Quick Start (Self-Hosting Guide)

For step-by-step signing, on-device speech model setup, and troubleshooting, see install.md.

Prerequisites

1. Devices: A Mac (macOS 15.0+, Apple Silicon recommended), an iPhone (iOS 17.0+), and an Apple Watch (watchOS 10.0+, or compatible Bluetooth earphones).
2. Build Tools: Xcode 16+ and xcodegen (brew install xcodegen).
3. AI Subscription (Choose Any): OpenAI Codex (ChatGPT Plus / Pro), Claude (Claude Pro / Max), Google Antigravity, or a custom OpenAI-compatible API.
4. Tailscale (Optional, for access outside home Wi-Fi): A free Tailscale account.

1. Clone & Configure Signing

git clone https://github.com/Miri-Labs/MoMo-Agent.git
cd MoMo-Agent
cp Config/Local.xcconfig.example Config/Local.xcconfig

Fill in your Apple Developer Team ID and bundle ID prefix in Config/Local.xcconfig:

MOMO_DEVELOPMENT_TEAM = YOUR_TEAM_ID
MOMO_BUNDLE_ID_PREFIX = com.yourname.momoagent

2. Launch the Mac Hub (MoMoAgentMac)

Or download the prebuilt MoMo-Agent-Mac-Apple-Silicon-*.zip directly from Releases.

xcodegen generate
xcodebuild -project MoMoAgent.xcodeproj -scheme MoMoAgentMac -configuration Release build

Open MoMo Agent on your Mac, sign in with Codex, Claude, Antigravity, or Custom API, and display your Pairing QR Code.

3. Install on iPhone & Apple Watch (MoMoAgent)

Open MoMoAgent.xcodeproj in Xcode, select the MoMoAgent scheme targeting your iPhone (which automatically installs the companion app on your paired Apple Watch), and press Run (⌘R). Scan the Mac QR code on your iPhone, enroll your voiceprint once, and start capturing.

Roadmap

• v0.0.1 — Self-Hosted Foundation: Apple Watch all-day capture, free offline transcription & speaker diarization, built-in Tailscale private mesh, and Mac agent hub.
• v0.3.x — Faster On-Device Speech & Earphone Capture: Apple Neural Engine speech models, Watch local Wi-Fi fast sync, and Bluetooth earphone recording.
• v0.4.x — Proactive Butler & Apple Ecosystem: Daily Morning Briefing, voiceprint-grounded owner-only task creation, and native Apple Reminders, Calendar & Notes integration.
• v0.5.3 — Self-Organizing Memory Vault, Life Illustrations & 3-Way Subscriptions: Multi-level Obsidian-style Markdown vault with audio proof and plain-language editing, illustrated daily life stamps & footprint map, plus seamless switching across Codex, Claude, and Antigravity.
• AirPods Seamless Route Switching: Smooth automatic audio routing when wearing AirPods.
• Two-Way External Vault & Workflow Sync: Direct two-way sync with external Obsidian vaults, Things 3, and local Git repositories.

Acknowledgments

MoMo Agent builds upon outstanding open-source work from DeepSeek SuperAgent, Sherpa-ONNX, SenseVoice, X-ASR, FunASR-Nano, Silero VAD, libfvad, Tailscale, and Hermes Agent.

License

MoMo Agent's original application source is licensed under AGPL-3.0-only. Third-party components and model weights retain their own licenses; the README device illustrations have separate noncommercial restrictions. See NOTICE.md.