Pipelyn

Read less. Know more — privately. Pipelyn turns any PDF, document, or video into clean text plus a deep, scored analysis — read by your own local model, with nothing leaving your machine. Drop in files or a whole folder, paste text, or fetch a YouTube URL; get a grounded judgment you can keep, sort, and search.

What it does

• Everything in, one pipeline — PDFs (born-digital and scanned, via fully-offline OCR), text, markdown, whole folders, pasted text, and YouTube links all normalize to the same clean, verbatim text.
• Deep triage, not a throwaway summary — your local model returns a grounded multi-paragraph description plus five reasoned 1–5 scores (depth, novelty, actionability, credibility, relevance) and an overall — a structured judgment you can sort and filter on.
• Reads the whole thing — map-reduce chunking lets a modest local model judge a book-length PDF on its full content, not its first page.
• Keeps everything — every result is saved to an on-disk library, content-hash deduped (a renamed file is still "already done"), and reopenable instantly with no re-run.

Private by architecture

No server, no account, no telemetry. The only network calls are to YouTube (for captions) and your own local Ollama — the OCR model even ships bundled, so scans are recognized with zero network. You bring your own free local model, so there is no subscription and no per-token bill. Privacy isn't a setting; it's the design.

Availability

Pipelyn is a cross-platform desktop app (Mac, Windows, Linux). Signed public installers are on the way — the download isn't ready yet, so check back shortly. You'll need Ollama installed and a capable-enough machine; deep triage of long documents takes minutes — it's thorough, not instant.