Find answers in the docs, browse the FAQ, or reach out directly.
Step-by-step guides for every feature — from adding content to doing a Feynman technique review.
Send us a message at [email protected]. We typically respond within 1–2 business days.
You need an API key for the in-app Chat panel, transcription, and cloud text-to-speech providers. Drag and drop still works with your existing Claude, ChatGPT, Gemini, Grok, or Edison Scientific subscription. In-app provider requests go directly from your Mac to the provider you choose, not through FeynmanLM servers.
All sources, study data, and generated artifacts are stored in your private iCloud container. The current app persists a local SQLite database plus source files there, synced by Apple. FeynmanLM servers do not store your source library. If you choose the hosted Cloudflare MCP relay, MCP traffic is proxied through FeynmanLM-operated infrastructure. See our Privacy Policy for full details.
Anthropic Claude (Claude Sonnet, Claude Opus), OpenAI (GPT-4o and newer), Groq (fast and cost-effective), and Google Gemini (for image generation). You can mix providers — for example, use Claude for answer evaluation and Groq for speech transcription.
Yes. FeynmanLM supports voice input with real-time transcription. Choose from on-device Apple Speech (completely private), OpenAI Whisper, Groq, and other providers. Configure your preferred provider in Settings → Speech To Text.
No. FeynmanLM is currently macOS only.
Download the PDF to your Mac and save it to ~/Library/Mobile Documents/iCloud~com~feynmanlm/Documents/FeynmanLM/Papers/. The app will find it automatically when you add the paper's URL. See Papers in the docs.
The MCP (Model Context Protocol) server lets Claude, ChatGPT, Gemini, Grok, or Edison Scientific access your source library in conversations. It can search sources, read source content, run Feynman reviews, and manage Schedule. It's optional. Tailscale or another tunnel you control is the privacy-first setup; the hosted Cloudflare relay is easier but proxies MCP traffic through FeynmanLM-operated infrastructure. See Setting Up the MCP Server.
Schedule progress is based on completed or linked Feynman reviews. Opening or listening to a source helps you study it, but the scheduled item is only done once a review session is saved for that source.