Built out of necessity
I’m Ashrya, an AI scientist. I read a lot of research papers and, I got tired of the same spiral: hit a paragraph full of jargons I have no freaking clue about => go to Claude & ask for clarification => then clarification of the clarification => and by the time I’d surfaced, where the hell am I?. So I built the reader I always wanted. When it was polished enough for me to use daily, it felt like it might be useful to someone else.
There’s nothing to sign up for, no subscription, no account. If you already pay for Claude, you have everything Fathom needs.
A new way to read a paper
Fathom asks you not to leave the document. The zoom frames the passage; you ask what you actually want to know. Nothing is predicted, nothing streams unprompted. One gesture to focus, one sentence to ask. Swipe back when you want to keep going.
It’s a reading app shaped around the way you already read, not an AI assistant bolted onto a PDF viewer.
What it feels like
Hold ⌘ and pinch on any passage. The page gives way to a full-screen lens anchored on exactly what you were looking at, with an input at the bottom. Type your question — the answer streams in. Pinch a phrase inside the lens to drill deeper. Swipe back, the way you came. Every lens persists across sessions: close the PDF, open it next month, click the amber marker on the page, and the thread you had is still there, exactly where you left it.
What makes it different
- The zoom is the explanation. No side panel, no context switch, no “AI assistant” icon. The gesture you’d already use to look closer is how you ask for help.
- Grounded in the paper itself. Claude is given a file-system index of
the paper —
content.md, per-figure PNGs, a digest — and navigates it withRead/Grep/Glob. No RAG, no embeddings, no similarity search. The paper is a filesystem; the AI is a shell. - Diagrams when they help. Architectures, pipelines, and relationships render as hand-drawn inline SVG. Never ASCII, never Mermaid.
- Durable across sessions. Every lens round-trips across app restarts: the viewport crop, the full thread, the exact prompt that was sent.
Install
Fathom’s primary install path is the terminal. One line — see the chip at the top of this page, or copy from here:
curl -fsSL https://raw.githubusercontent.com/ashryaagr/Fathom/main/install.sh | bash
The script extracts Fathom to /Applications, clears the
com.apple.quarantine xattr so Gatekeeper doesn’t prompt, ad-hoc
re-signs, and drops a fathom launcher at ~/.local/bin/fathom
so fathom, fathom paper.pdf, and fathom update work from any
shell. Then it launches Fathom — you land on the welcome screen in
one motion. Full walkthrough in the
install guide.
Prefer a drag-to-Applications install? See the Mac DMG section of the install guide — the DMG download link + the one-time Gatekeeper approval walkthrough live there.
Your data stays yours
Fathom runs entirely on your machine. No telemetry. No analytics. No accounts. No server ever sees your PDFs, your explanations, or your conversations with Claude. The only network calls are your own Claude Code CLI talking to Anthropic on your behalf, and the app’s auto-updater checking GitHub for new Fathom releases.
Every paper’s index and chat history lives under
~/Library/Application Support/Fathom/. Delete that folder any time
to wipe all of Fathom’s state; your PDFs themselves are untouched.
Free and open source
Fathom is MIT-licensed and built in the open.
- Source →
- Releases →
- Methodology → — how Fathom actually works under the hood: extraction pipeline, grounding strategy, why not RAG.
- Design principles → — the rules Fathom was built on. Read before proposing changes.
- Report a bug →
There’s no roadmap document because the roadmap is whatever the paper I’m reading this week demands. If you’re using Fathom on something it handles badly, opening an issue with the PDF attached is the most direct way to shape what ships next.