Open-source meeting recorder with local transcription and AI notes
AI: publik API or your key
publik verified: the founder reviewed this exact version for malware and approved it
Manual download To get a new version, download it again from publik.

Prices are for one typical use: one request of about 1,500 words sent and 375 words back. A higher quality score is better.
Not checked yet: publik has not read whether Oatmeal can run its AI on your computer.
| Model | Quality | Per 1,000 uses |
|---|---|---|
| On your computer (Ollama, 4-bit download size) | ||
| Granite 4.2 3B2.2 GB | 9 | $0 |
| Phi-4 Mini2.5 GB | 6 | $0 |
| Llama 3.1 8B4.9 GB | 7 | $0 |
| gpt-oss 20B14 GB | 9 | $0 |
| Gemma 4 31B20 GB | 19* | $0 |
| Qwen3.5 35B-A3B24 GB | 19* | $0 |
| publik API | ||
| publik-fastGLM-5.3 Flash | 42 | $1.00 |
| publik-balancedMiMo-V2.6-Pro · Oatmeal | 46 | $10.00 |
| publik-smartGPT-6 Sol | 48 | $18.00 |
Quality: Artificial Analysis Intelligence Index v4.3.2, read 2026-09-22 (publik API models 2026-09-25); * = estimated by Artificial Analysis. Sizes: the Ollama library, read 2026-09-22.
An open-source meeting notebook with local transcription and your choice of AI.
Listen, jot down a thought, and leave with the details that matter.
Run it today · Watch the demo · On Publik ↗ · Road to release
The recorder that ships on main today, transcribing a scripted sample meeting with local Whisper.
[!NOTE] What you get from
maintoday: a local browser recorder (Whisper in your browser, You/Room lanes, Markdown transcripts) plus your coding agent — Claude Code, Cursor or Codex — writing the notes. That's what installs from this repo and from the Publik listing. It opens with Sign in with publik (free, once per computer); the recorder itself calls no AI service.Still in preview: the standalone desktop app with built-in AI and the Publik API option lives in draft PR #2. Its Publik path is implemented and tested against a mocked service, but a live paid generation hasn't been verified yet, and there are no signed installers or GitHub releases.
Your meeting is the main event. Oatmeal keeps a local transcript while you take the few notes that matter, then helps you find the decisions, commitments, and next steps.
Capture the conversation. Record your microphone and optional meeting audio. Whisper transcribes on your computer. No bot joins the call.
Add your perspective. Write in a clean notebook while the conversation flows. Notes save automatically.
Keep what matters. Make a summary, ask about the meeting, search your library, or export everything as Markdown.
On main, your coding agent does the summarizing and searching (/writeup, /recall); the desktop preview adds a built-in notebook and AI.
The browser recorder on main, recording a sample meeting. Whisper transcribes it in the browser, then Claude Code writes the notes from the transcript file.

▶ Watch the 35-second demo (MP4) — record → live transcript → saved Markdown → notes.
Real run of main in Chrome. The meeting is a scripted sample read by macOS text-to-speech into a virtual mic (You) and a simulated shared tab (Room). Transcription is real local Whisper base, mistakes included; the notes were written by Claude Code from that transcript with /writeup. The GIF and the recording part of the video play at 2×. Media notes →
| Ready to record | Live transcript |
|---|---|
![]() | ![]() |
| Saved as Markdown | Notes written by your agent |
![]() | ![]() |
Not on main yet. Example meeting with fictional content.

Desktop preview UI walkthrough with fictional notes, transcript, and a prepared summary. Generation and recording are not shown.
Audio is transcribed locally. In the desktop preview, choose where the text goes when you ask for AI help. On main, the notes are written by whichever coding agent you already use.
Claude |
OpenRouter |
Ollama |
Local models |
| Option | How it works |
|---|---|
| On this computer | Download Whisper for speech and Qwen for text. Work offline after the downloads, with no account or usage bill. Small local text models select original meeting excerpts to preserve facts. |
| ChatGPT subscription | Sign in through the official Codex runtime. Your plan’s Codex limits apply. |
| Claude subscription | Sign in through the official Claude Agent SDK. Eligibility and usage follow your Claude plan. |
| Your API key | Connect OpenAI, Claude, Grok, or OpenRouter. Keys saved in Settings use operating-system encryption. |
| Publik API | Choose Fast, Balanced, or Smart. Link an account and use a shared Publik balance; the app shows costs and account links. A fresh install starts at $0.00. Implemented in PR #2 and tested with a mocked service; a live generation is not verified yet. |
| Ollama | Connect to an existing local Ollama server and choose your model. |
Cloud requests require confirmation. Raw audio stays on your computer. Cloud providers receive meeting text only when you ask them for AI help. Read the privacy and account details →
With Node.js 20+ and Chrome or Edge:
git clone https://github.com/VedSoni-dev/oatmeal.git
cd oatmeal
npm ci
npm start # recorder at http://localhost:4123
meetings/ (or OATMEAL_MEETINGS_DIR)./writeup — or ask Cursor/Codex to follow SKILL.md — to turn the transcript into notes with decisions and action items. That step uses your agent's model and account.node scripts/install-autostart.mjs keeps the recorder running in the background. Agent setup guide →
Sign in with publik and usage counting. Oatmeal is listed on publik, and every publik app opens with Sign in with publik (capture/publik.mjs, publik's official module). Once you're signed in, the recorder sends publik two small events so it can count weekly users: app_opened when the server starts and active_session every 30 minutes while it runs. They carry the app name and version, your OS and CPU type, and a random install ID. No meeting audio, transcripts, titles, file names or prompts are sent. Turn it off with DO_NOT_TRACK=1 or PUBLIK_TELEMETRY=off, or by setting "usage": false in publik's consent.json (details). The sign-in key is stored in publik's per-app credential file (mode 0600), never in this repo. PUBLIK_SELF_TEST=1 npm start prints publik self-test: ok oatmeal-vedsoni-dev <version> and exits.
The desktop app is a development preview in PR #2. Apple Silicon development builds have been tested. Windows and Intel Mac packaging are configured, with platform validation still pending. Signed public installers are not available yet.
With Node.js 22+, run:
git clone --branch codex/desktop-app https://github.com/VedSoni-dev/oatmeal.git
cd oatmeal
npm ci
npm start
This checks out the preview branch, not main. Oatmeal opens in its own desktop window. No coding agent or separate local model server is needed.
Run on the matching operating system and architecture:
npm run build:mac # .app, .dmg, .zip — macOS 14.2+
npm run build:win # NSIS installer — Windows 10/11 x64
Local builds are development artifacts. See signing and release setup before distributing them. Real subscription sign-in, platform audio permissions, and long meetings still need release validation; verification status records what was actually tested.
The original browser recorder and Markdown workflow are still here. Work with meeting files in Cursor, Claude Code, Codex, or another MCP client.
Cursor MCP + Markdown |
Claude Code Skills + MCP |
Codex MCP + Markdown |
The MCP server exposes list_meetings, search_meetings, and get_meeting for Markdown files. Export desktop meetings into the folder you configure for MCP; it does not automatically read the desktop database. Agent setup guide →
Meeting files are not encrypted by the app. System-audio capture depends on the OS, permissions, and chosen source. You/Room separates audio lanes, not individual remote speakers.
Bug reports, thoughtful design feedback, and pull requests are welcome. Include your OS, provider, and steps to reproduce; keep meeting content and credentials out of public issues.
On the desktop preview branch:
npm test
npm run test:desktop
Desktop guide · Architecture · Release checklist · Report an issue
MIT © Vedant Soni. An independent Granola alternative. WhimprFlow inspired the platform and release patterns. Product names and logos identify the tools and services described; they belong to their respective owners and do not imply endorsement. Asset credits