For PDFs
AI for PDFs — even the scanned ones.
Sidenote is a Chrome extension that reads any PDF you open in the browser — including scanned, image-only PDFs via OCR. Summarise it, chat with it, ask it anything. Every answer cites the exact passage and scrolls the document straight to it. You read and question the PDF — you don't just convert it.
The problem
The answer is in the PDF. Reading all of it isn't the plan.
PDFs are where the important things live — contracts, reports, papers, datasheets, policies — and they're the format least built to be questioned. Search finds a word, not an answer. And the moment a document is scanned, even search gives up: it's a picture, not text.
Sidenote reads the PDF for you — running OCR on scanned pages when it needs to — and answers in your browser, on the file you already have open, with a citation you can click to scroll straight to the source.
How it works
How Sidenote works on a PDF.
Working through a stack of files? Bundle them into a Collection and query several PDFs at once — citations name the source document and passage.
Use cases
What people ask a PDF — and how Sidenote answers.
- “What does this contract say about termination?”Sidenote reads the PDF you've opened and quotes the exact clause — then scrolls you to it so you can read the surrounding context yourself.
- “Summarise this scanned report.”A scanned, image-only PDF isn't searchable text. Sidenote runs OCR first, then summarises — so even a photographed document becomes answerable.
- “Explain this research paper in plain English.”Dense methods sections become a short, cited explanation. Click any claim to jump to the figure or paragraph it came from.
- “What are the key numbers in this datasheet?”Pulls the spec or figure you asked for and points at the row it sits in — no scrolling through forty pages to find it.
- “Does this policy actually cover my case?”Reads the PDF and quotes the clause that applies. If nothing in the document supports an answer, Sidenote says so rather than guessing.
- “Give me the TL;DR of this 80-page PDF.”A long PDF becomes a short, cited summary — every line links back to the page and passage it was drawn from.
Scanned PDFs
Yes, it reads scanned PDFs too.
A scanned PDF is a stack of page images with no machine-readable text underneath — which is why you can't select it, search it or paste it into a chatbot. Sidenote runs OCR to lift the text out of those images first, then summarises and answers over it like any other document — citations included.
Photographed contracts, old reports, faxed forms. If you can read it with your eyes, Sidenote can usually OCR it and answer questions about it. And because every answer is checked against the extracted text, anything the document doesn't actually support gets dropped before it reaches you.
Comparedvs Smallpdf / iLovePDF / AdobeBrilliant for converting, compressing, merging and signing — all of which transform the file. Sidenote leaves the file alone and answers questions about its contents instead, with a citation that scrolls to the exact passage, scanned documents included. vs pasting into ChatGPTCopy-paste loses the layout and the source — and scanned PDFs won't copy at all. Sidenote ingests the document in place (OCR and all), checks every answer against the retrieved passages, and links each claim back to where it came from.
Read and question — not just convert.
Most PDF tools change the file. Sidenote reads it. That's a different job — and the one you actually need when the question is “what does this say?”
Just want the gist of one file? Sidenote turns a long PDF into a short, cited brief — and every claim is tied to a citation you can verify in one click.
Private by design
Your documents stay yours.
A PDF is only ever read once you open it or upload it — there's no background crawl and no copy of every file sitting on someone else's server. Your documents are used to answer your own questions and nothing else.
Stored in your account, isolated per user. Document text lives behind row-level security in our UK (eu-west-2) Supabase region — no two accounts can read each other's content. Anthropic and Voyage AI both run with no-training defaults on the API tiers we use; we never fine-tune models on user content.
FAQ
Common questions about Sidenote for PDFs.
Can Sidenote read a scanned PDF?
Yes. A scanned PDF is really just images of pages with no underlying text, so ordinary tools can't read it. Sidenote runs optical character recognition (OCR) on the scanned pages first to extract the text, then summarises and answers questions over it — with citations back to the page the text came from.
Do I have to upload the PDF anywhere?
Not usually. If the PDF is open in a browser tab, Sidenote reads it in place from the side panel — no upload step. For files on your computer you can also upload them to the web app. Either way, you're reading and questioning the document, not converting it.
How is this different from Smallpdf, iLovePDF or Adobe?
Those tools mostly convert, compress, merge or sign PDFs — they change the file. Sidenote doesn't touch the file; it reads it and answers questions about it. You ask “what does this say about X?” and get a grounded answer with a citation that scrolls to the exact passage, including in scanned documents via OCR.
Does every answer really cite the source?
Yes — that's the core of it. Each answer is grounded in quoted passages from the PDF, and any claim the retrieved text doesn't support is dropped server-side before you ever see it. Click a citation and Sidenote scrolls the document to the exact passage and highlights it.
Can I ask questions across several PDFs at once?
Yes. Bundle related PDFs into a Collection and query the whole set in one go — citations name which document and passage each answer came from. Useful for a stack of papers, a set of contracts, or a folder of reports.
Is my PDF used to train AI models?
No. Anthropic (the model provider) and Voyage AI (the embedding provider) both run with no-training defaults on the API tiers Sidenote uses, and Sidenote never fine-tunes models on user content. Your documents are used only to answer your own questions, stored in a UK (eu-west-2) region with row-level security isolating every account.
Try it on your next PDF
Make any PDF answer back.
Add Sidenote to Chrome, open a PDF — scanned or not — and ask it anything. Every answer scrolls you to the source. Free tier forever, 7-day Pro trial — no card required.
Reads PDFs in the browser · Scanned PDFs via OCR · Upload optional