Turn any web page, PDF, or EPUB into a study set

Bring your own source

Novus Learn isn't limited to Wikipedia. Paste a URL to read and break down a web page, point it at a site to index, study a journal article, or upload a document. Whatever the source, you get the same source-grounded study views.

This is the path that makes the tool useful for work rather than only for browsing. The reading you actually have to do is rarely an encyclopedia article — it's a specification, a textbook chapter, a paper someone sent you, a policy document with a deadline attached.

Those documents share a property that makes them miserable to study: they were written to be complete and correct, not to be learned from. The ordering serves the argument rather than the reader, definitions arrive after the terms that need them, and the important sentence is often in the middle of a paragraph about something else. Pulling the claims out and letting you test yourself against them is a way of reading such a document actively rather than hoping a second pass will stick.

Formats that just work

Uploads are read by a format-specific extractor, so the list is concrete:

• Plain text and Markdown, read directly.
• PDF, parsed page by page with a self-hosted PDF engine.
• EPUB, unpacked and read chapter by chapter.
• Word (.docx) and PowerPoint (.pptx), unzipped and extracted from their XML.
• Caption files (.vtt and .srt) for anything you already have a transcript of.
• A .zip of source code, for reading a codebase rather than a document.

Built for things that are actually long

A book-length upload is broken down chapter by chapter so you can study it in pieces instead of facing one giant wall of text. Extraction runs per chapter, with progress reported as it goes, and the resulting project keeps that structure rather than flattening it.

The limits are deliberately generous but real: around ten megabytes for plain text and forty for the binary formats, with an eight-million-character ceiling on what gets kept. Those numbers exist because everything runs in your browser tab — the ceiling is what keeps a huge file from taking the page down with it.

All of the parsing happens behind a worker, with safeguards against malformed or oversized files, so a big upload never freezes the page or gets shipped off to a server.

Every breakdown is a little different

A how-to guide, a news report, and a research paper don't read the same way, so Novus Learn leads with the views that fit each one — a step-by-step walkthrough for a guide, a quick digest for news, a parts breakdown with sources kept close for a paper.

That decision is made from the page itself rather than from a setting you have to find: the kind of page, the presence of a real procedure, the way sections are structured. Get a documentation page and you'll be offered a how-to; get an essay and you'll be offered a breakdown into its parts.

What upload quality depends on

The honest variable is the file. A born-digital PDF with a real text layer extracts cleanly. A scanned PDF is a stack of images with no text to find, and Novus Learn will not invent it — there is no OCR step. A heavily-designed export where the reading order is scrambled will produce sentences in the wrong order, because that is genuinely what the file says.

Layout is the other thing worth expecting. Tables, footnotes, figure captions and multi-column layouts survive unevenly across formats. EPUB and DOCX tend to extract most faithfully because they carry real document structure; PDF is the format most likely to need a skim before you trust the breakdown.

Which tool for which source

There are several entry points and it isn't always obvious which one you want:

• One web page you want to understand — the URL reader.
• A whole documentation site or a set of related pages — the site indexer.
• A research paper, where the sections and citations matter — the journals tool.
• A file on your machine, in any of the supported formats — upload.
• A book or anything long enough to need chapters — the textbook breakdown.
• A lecture or talk you only have as video or audio — the video tool.

A quick way to start

If you want to see the difference in a minute: take a documentation page you half-understand, paste it into the URL reader, and open the breakdown. The value isn't the summary — it's that every line in it is a link back into the page you were struggling with, so the parts you skipped become obvious.

The failure mode this fixes is a familiar one. You read something dense, it sort of makes sense, and you move on with a false impression of having understood it. Being handed the page's actual claims as separate, checkable items makes the gap between “I read this” and “I follow this” impossible to miss — which is uncomfortable, and exactly the point.