For people who read for a living
AI gives you answers.
SCL is your shortcut to understanding.
Your papers — organised, searchable and explained in your first session, without moving a single file. You keep your edge: a local, loyal AI over your own library, not a datacenter AI over your uploads. Local by default; anonymised only when you ask for a frontier model.
Join the first 100 See it work Private beta. The installer isn't code-signed yet — we explain exactly what that means.
990 KB · free · works offline · your files stay on your machine
“My PDFs are a mess.”
Point SCL at your drive. It shows everything you already have in topic folders — organised by what the papers are about, not what their filenames say.
Nothing moves. Not one file is relocated, renamed or touched.
- ▸ Condensed matter (412)
- ▸ Superconductivity (96)
- ▸ Topological phases (74)
- ▸ Machine learning (388)
- ▸ Methods & instrumentation (213)
“I can’t find anything inside them.”
One search box, across the full text of every publication you hold. Contents, not filenames. The paper you half-remember from 2019 is findable by the sentence you half-remember.
- 🔍 “flux pinning in layered” — 23 documents
- — ranked by relevance, opened at the matching passage
“I can’t possibly read it all.”
Every paper gets a faithful paraphrase at the depth you choose — expert, layman, or novice — and a Q&A panel: purpose, method, biases, what it actually claims. You stay the expert; the AI does the triage.
3,000,000 papers arrive already prepared — most of arXiv. If your folder holds one of them, its paraphrases, links and analysis appear instantly. Anything else, you select for processing and are told when it is ready.
- Expert · Layman · Novice
- “The authors demonstrate that vortex lattices in …”
“I don’t know what I already have.”
SCL cross-links everything you hold. Open a paper and see the others on your own drive that bear on it — including the one you forgot you downloaded three years ago. Real links between ideas, not keyword matches.
- ↔ Kim 2021 — same phenomenon, different measurement
- ↔ Aoki 2017 — the theoretical framework this result tests
The arithmetic nobody says out loud
Publication volume grows every year. Your reading capacity does not. That gap is not a personal failing — it is arithmetic. See yours:
Approximate volumes from arXiv’s public submission statistics (preprints only — journal-only publications are not counted, so the real number is higher).
“…but using AI means giving my work away.”
Every AI answer costs you something. What you hand over to get it is your Privacy Gap.
Your documents, your questions and your line of thinking — on someone else’s machine, under a policy that can change without you.
Content is stripped before anything goes out, so you can reach a frontier model when you need one. Less leaves — but something still leaves. We won’t pretend otherwise.
Your documents, your questions and your answers stay put. The only setting where the gap is actually zero — and our default. Check it yourself.
You are no longer choosing between a good answer and a private one. How the two modes work, and what we will not claim →
What it costs
Free keeps every feature. Paying buys speed and convenience — never access, never privacy. See the price list →
Who it’s for
Researchers, engineers, graduate students, independent experts, inventors. If your livelihood runs on actually knowing things, this was built for you.
Use the giants; don’t teach them your experience.
Ten minutes to a working library
A 990 KB installer. Point it at one folder. Watch your own papers come back organised, searchable and explained — without a single file moving.
Join the first 100 Private beta. The installer isn't code-signed yet — we explain exactly what that means.