Last updated 10 min read
Key takeaways
A PDF is built for printing, not for learning. A 60-page course reader, a scanned textbook chapter or a 30-page research report arrives as one long scroll with nobody to ask when a paragraph stops making sense. You can search it for a word, but you cannot ask it why the author drew a conclusion, what a column in a table means, or how page 12 connects to page 40. Most students fall back on highlighting, which feels productive and leaves very little behind.
An AI tutor that can read the file changes what you can do with it. Instead of a fixed document, you have something you can question: point at a paragraph and ask what it claims, ask for the steps behind a derivation, or ask to be tested on the section you just finished. That is genuinely useful, and it is also where the risk sits. Chatting with a PDF makes it very easy to stop reading it, and a tidy summary of a document you have not read is not the same as knowing what is in it.
Not every PDF contains text. Some are digital documents with a real text layer; others are photographs of pages, produced by a scanner or a phone camera, where every page is a picture. An AI can only work reliably with words it can extract, so spend thirty seconds checking before you upload. Open the file, try to select a sentence with your cursor and paste it somewhere. If clean text appears, the file is readable. If nothing selects, or the pasted text comes out garbled, the page is an image or the text layer is poor.
Scanned files can usually be fixed with optical character recognition (OCR), which many PDF readers and scanning apps can run. After OCR, spot-check a page with equations, footnotes or a table, because that is where recognition errors cluster. A misread minus sign or a dropped decimal point will be repeated back to you with complete confidence.
Also note how the document numbers its pages. A chapter scanned from page 213 of a book will show 213 in the margin but be page 1 of the PDF. Decide which numbering you will use for page references and say so in your first message, or you will waste time hunting for pages that do not match.
This routine works for a textbook chapter, a lecture handout or a research paper, in a session of about an hour. Steps five and six do most of the work: they turn the PDF from something you read into something you can retrieve, so do not skip them.
The questions that help most change as you move through a document. Early on you want orientation; while reading you want explanations of specific passages; afterwards you want to be tested. The table gives examples for a typical study PDF.
Notice what the table below leaves out: “summarize this PDF.” A summary has its place as a preview before reading or a recap afterwards, and the guide to types of AI study tools explains where summarizers fit alongside tutors and flashcard apps. As a replacement for the reading itself, it removes the effort that makes the material stick.
| Stage | Good questions to ask | What you get |
|---|---|---|
| Before reading | “What question is this document trying to answer?” “Which sections define the key terms?” | A purpose and a map, so details have somewhere to attach |
| While reading | “On page 14, what does ‘elasticity’ mean in this context?” “Walk me through the worked example on page 22, one step at a time.” | An explanation of the exact passage that stopped you |
| Right after | “Quiz me on sections 2 and 3, one question at a time, and wait for my answer.” | Retrieval practice on what you just read |
| Days later | “Ask me five mixed questions from the whole document, hardest first.” | A check on what has lasted, not what you saw an hour ago |
Even an AI reading your own file can get things wrong. It can blend a claim from page 9 with one from page 31, misstate a figure from a table, or fill a gap with something plausible that the document never says. The habit that protects you is simple: ask for the page with every substantive answer, and open that page to confirm the claim is there in roughly those words. If the tutor cannot point to a page, treat the answer as unverified.
Be most careful with numbers, names, dates, direct quotations and anything that sounds like a definition you might reproduce in an exam. These are the details that are easy to state confidently and costly to get wrong. The guide to catching AI hallucinations covers the warning signs in more depth.
Typed chat suits a PDF well because you can quote phrases and scroll back through answers. Some confusions clear faster by talking, though. If you have asked about the same derivation twice and it still has not clicked, or you want someone to follow your reasoning while you explain a chapter aloud, a voice or video call with the tutor moves faster than another round of typing. Explaining out loud also exposes gaps that writing hides, because you cannot quietly skip the step you do not understand.
A practical pattern is to read and question in chat, then end the session with a short call: explain the main argument of what you read without the file in front of you, and let the tutor ask follow-up questions about the weak spots. Keep the chat history for anything you will want to look up later.
On Cavua, your Work space is where your own documents go. You can upload PDFs, Word files and more, chat about them, research across several at once with answers that point to the page, ask the tutor to “Quiz me,” and export a session when you are done so your notes leave with you. You can also video or voice call the tutor about your documents, which suits the explain-it-back step above.
How much of this is included depends on your membership plan, and each plan states its allowance on the pricing page. If you are taking a Cavua course rather than studying your own file, the course's tutor already answers from that course's syllabus, lessons and tests, so there is nothing to upload.
Most problems with studying a PDF through AI come from letting the conversation replace the reading. Watch for the mistakes listed below.
If your PDF is a textbook chapter, the wider routine in how to study a textbook and remember it pairs well with this one, and prompts for studying with AI shows how to ask for practice questions that get harder as you improve.
AI prompts for studying that teach rather than tell: a five-part template, weak vs strong examples, and copy-ready prompts for understanding and practice.
AI tutor vs ChatGPT for studying: what each knows before you ask, where each is strongest, and how to use both without getting mixed messages.
How to study with AI the right way: attempt first, use it to question and correct you, check its answers, and measure what you can recall on your own.
Is using AI cheating? It depends on your course policy and what the work assesses. See where the line usually falls, how to check, and how to disclose AI use.
What a video call with an AI tutor is really like: what you see, how the conversation flows, how it compares with a human tutor, and how to prepare for it.
An AI tutor answers questions about one specific course in real time, by video call, voice call or chat. Here is how course-grounded tutoring works.
Yes. Many AI study tools, including Cavua's Work space, can read an uploaded PDF and ask you questions about it. To get the most from this, read the section first, then ask for questions one at a time and answer each before seeing the next. Ask for a page reference with every answer so you can check the source, and come back a few days later for a mixed quiz on the whole document.
The most common reason is that the PDF is a scan, so each page is an image rather than text. Try selecting a sentence; if nothing highlights, run optical character recognition (OCR) in a PDF reader or scanning app before uploading. Password protection, very large files and unusual layouts can also cause problems, so test with a short section before relying on the tool for a whole chapter.
Ask it to name the page the answer comes from, then open that page and check. If the claim is there in roughly those words, you can rely on it. If the tutor cannot give a page, or the page says something different, treat the answer as unverified. Be especially careful with figures, dates, quotations and definitions, since those are easy to state confidently and get wrong.
Some tools can work across several documents in one conversation, which helps when a topic is spread over a textbook chapter, lecture slides and a paper. Name the files clearly and ask which document and page each answer draws on. That way you can tell whether two sources genuinely agree or whether the tool has blended them into one claim neither actually makes.
That depends on your institution's rules and the terms the material was shared under, so check your course policy and the tool's terms before uploading. Using AI to understand material you legitimately have access to is generally treated differently from submitting AI-written work as your own. This is general information, not legal advice; ask your instructor if you are unsure.
Every Cavua course comes with a 24/7 tutor you can video call, speak to or chat with — on the material you are actually studying.
