Libro VS CypherScan

CypherScan's answer:

Most tools that chat with your documents keep a copy on a server — and that copy is exactly what lets you reopen the conversation next week, so the convenience and the storage are the same fact. CypherScan makes the other trade. Word, text and spreadsheet files are parsed in the browser and never leave the device; PDFs and images are sent, and the interface says so before you press the button rather than burying it in a privacy policy. Nothing is stored on the server and nothing trains a model. The cost is real: there is no history, so you open the file again to ask again.

CypherScan's answer:

Mostly for what it does not do. There is no document store to breach, no content log for staff to read, and no training use of what you send. It also starts without friction — three files a week are free, with no account and no card, and signing in later is a link in an email rather than a password to lose. If you want a searchable library of everything you have ever uploaded, a storage-based tool is genuinely the better choice. If you want an answer without leaving the document behind, this is built for that.

CypherScan's answer:

People handling documents they would rather not leave on someone else's server: lawyers and accountants reading contracts and statements, freelancers and small businesses working through invoices, researchers and students going through papers, and anyone reviewing a file that contains someone else's personal data. The interface is in five languages, so a large part of the audience is outside English-speaking markets.

CypherScan's answer:

It started from an ordinary irritation: reading forty pages to find one clause, and noticing that every tool offering to do it for you wanted a permanent copy first. The interesting question was how much of the work could happen on the reader's own machine. Word, text and spreadsheet files turned out to need nothing but the browser. PDFs and images still have to be sent, so the product says that plainly instead of hiding it — a claim you can verify is worth more than one you have to trust.

CypherScan's answer:

A static site of plain HTML, CSS and JavaScript with no framework, generated as separate pages per language and served on Vercel with serverless functions. Parsing happens client-side: pdf.js for PDFs, Mammoth for Word, SheetJS for spreadsheets, and DuckDB-WASM to run real SQL over data files inside the browser. Analysis goes through LLM API. Accounts and quota live in Supabase (Postgres with row-level security, passwordless magic-link sign-in). Payments run through Paddle as merchant of record.