Open-source, local-first AI dictation app that turns speech into text on macOS and Windows, working offline with Whisper.
- Stars
- 129
- Forks
- 19
- Open issues
- 7
- Last commit
- 13 days ago
What is it?
Prismical is an open-source AI note taker for macOS and Windows that transcribes meetings, lectures and voice notes. It captures system audio in the background, so no bot needs to join the call, and works with apps such as Zoom, Google Meet, Microsoft Teams, Slack and WebEx.
Transcription can run on-device with whisper.cpp, or go through cloud or local models: OpenAI, Anthropic, OpenRouter, any OpenAI-compatible endpoint, or Ollama. The AI turns transcripts into structured notes with key decisions and action items, and includes built-in skills such as Enhance, Cleanup and Ask, plus custom skills.
The desktop app is built with Electron, React and TypeScript, with SQLite storage and full-text search across meetings and notes. An optional Prismical cloud adds cross-device sync, sharing and calendar features, and iOS and Android apps are available separately.
Key features
- 01Captures system audio in the background without a bot joining the call
- 02On-device transcription with whisper.cpp or cloud and local model providers
- 03AI-generated summaries with key decisions and action items
- 04Voice notes that AI structures and organizes
- 05Full-text search across meetings, notes and transcripts
- 06MCP server to connect notes to Claude, ChatGPT and Gemini (requires a Prismical account)
Use cases
- Recording and summarizing video calls without inviting a bot into the meeting
- Taking structured notes from university lectures or talks
- Capturing quick voice memos and having AI tidy them up
- Running transcription fully on-device with local models for privacy-sensitive work
Quick start
- 1
Download the desktop installer for your platform from the releases page
- 2
Clone the repository with the whisper.cpp submodule
git clone --recurse-submodules https://github.com/amicalhq/prismical.git - 3
Enable corepack and install dependencies, then build the workspace
corepack enable && pnpm install && pnpm build - 4
Run the desktop app in development mode
cd apps/desktop && pnpm dev
Things to know
- Requires Node.js 24, pnpm, CMake and Xcode or Visual Studio build tools to build from source.
- Local transcription needs macOS 14 or later, and system audio capture needs macOS 14.2 or later.
- Desktop builds are only for macOS and Windows x64.
In collections
Topics
FAQ
- What is prismical?
- Open-source desktop AI note taker that transcribes meetings and voice notes without a bot joining the call, using local or cloud AI. Prismical is an open-source AI note taker for macOS and Windows that transcribes meetings, lectures and voice notes. It captures system audio in the background, so no bot needs to join the call, and works with apps such as Zoom, Google Meet, Microsoft Teams, Slack and WebEx.
- Is prismical free to use? What license does it have?
- prismical is open source under the MIT license. Check the license terms before using it in commercial products.
- Is prismical still maintained?
- Actively maintained. The last commit to prismical was on Oct 26, 2026.
- What can I use prismical for?
- Recording and summarizing video calls without inviting a bot into the meeting Taking structured notes from university lectures or talks Capturing quick voice memos and having AI tidy them up Running transcription fully on-device with local models for privacy-sensitive work
- How do I install prismical?
- Download the desktop installer for your platform from the releases page Clone the repository with the whisper.cpp submodule: git clone --recurse-submodules https://github.com/amicalhq/prismical.git Enable corepack and install dependencies, then build the workspace: corepack enable && pnpm install && pnpm build Run the desktop app in development mode: cd apps/desktop && pnpm dev
- Where can I get prismical?
- The source code and installation instructions are on GitHub: https://github.com/amicalhq/prismical
Summary generated by AI from the project's README. GitHub stats refresh daily.
Related projects
Privacy-first AI meeting assistant that records, transcribes and summarizes meetings locally on macOS and Windows, with optional Ollama or cloud LLM summaries.
Open-source desktop dictation app that turns speech into text at your cursor, with local Whisper/Parakeet models or cloud providers.
Open-source, local-first meeting and voice memo recorder that transcribes on-device into Markdown files that AI agents like Claude Code can search via MCP.
Open-source macOS meeting-notes app that records mic and system audio without a bot, transcribes locally or via cloud APIs, and writes summaries from your notes plus the transcript.
A macOS meeting note-taker that transcribes both sides of a call locally and surfaces talking points from your own notes.