"Chatting with a PDF" means asking questions in plain English and getting answers grounded in that specific document, instead of reading the whole thing yourself. There are three practical ways to do it today, in increasing order of convenience and decreasing order of manual effort.
Method 1: Copy-paste into a general chatbot
This works with any AI chatbot you already have access to (ChatGPT, Claude, Gemini) and costs nothing extra.
- Open the PDF and select the text you want to ask about — or all of it, if the document is short.
- Copy it (⌘C).
- Paste it into the chatbot with your question, e.g. "Summarize this in three bullet points" or "What does this say about termination clauses?"
- Ask follow-up questions in the same conversation — the pasted text stays in context.
Where this breaks down: most chatbots have a context limit — a maximum amount of text they can hold in one conversation. A 5-page PDF pastes in fine. A 200-page research paper or contract won't fit, and if it's a scanned PDF, there's no text to select at all — you'd need OCR first (see how OCR works).
Method 2: A browser-based AI PDF tool
These upload your file and run the same extract-chunk-embed-retrieve pipeline described in how AI understands PDFs automatically, so you don't have to copy anything.
- Go to the tool's site and upload your PDF (this is the step to pay attention to for anything sensitive — see the FAQ below).
- Wait for it to process — for a long document, this can take anywhere from a few seconds to a minute, since it's chunking the whole thing and generating embeddings before you can ask anything.
- Ask your question in the chat box. Good tools cite which page or section the answer came from; treat tools that don't cite sources with more skepticism.
Where this breaks down: every browser tool requires an upload step, even for documents you'd normally just glance at in Preview. That's an extra step, an extra privacy consideration, and often a paywall after a handful of free uploads.
Method 3: A native Mac app with built-in AI
- Open the PDF directly in the app — no upload dialog, since it's already on your Mac.
- Open the AI panel (in PDF Go, this is the "Ask" panel) and type your question.
- The app locates the relevant part of the document and answers with page references.
Where this wins: you skip the upload step for basic viewing, and the app can point you straight to the page an answer came from, since it already has the document open rather than working from a copy-paste blob with no page boundaries.
Which method to actually use
- Short document, one-off question, don't already use an AI-PDF tool: copy-paste into whatever chatbot you already have.
- Long or scanned document, occasional use: a browser tool, mainly for the automatic chunking and page citations.
- PDFs are a daily part of your work — research papers, contracts, reports: a native app pays for itself in the upload steps alone, plus you get on-device markup (highlighting, drawing, signing) as part of the same app instead of switching tools.
A concrete example
Say you have a 40-page research paper and want to know what the methodology section says about sample size, without reading the whole thing.
- Copy-paste: find and copy the methodology section manually, paste it with your question. Fast if you already know roughly where it is.
- Browser tool: upload the whole paper, ask "what sample size did the methodology use," get an answer with a page citation, without knowing where to look yourself.
- Native app: same as the browser tool, but the paper was already open in your reader, so there's no upload step and you can click straight to the cited page.
All three get you a correct answer for a well-structured, OCR-clean document. The difference shows up on messier inputs — scanned pages, huge files, documents where you don't already know the rough location of what you're looking for.