Dragging files in for help

TL;DR:

• Dropping a file in gives the assistant the file's contents to read, so you can ask about page 3 instead of retyping it.

• Typed PDFs carry real text. Scanned pages are pictures, and the words must be recognized first, which is less reliable.

• Treat a dropped file like an email attachment: only share what you are allowed and willing to share.

Your teacher hands out a two-page worksheet as a PDF. Question 4 reads: "A cart rolls down a ramp and gains 3 m/s every second. How fast is it after 5 s?" You could retype it into a chat box, or you could drag the file onto the assistant and say "explain question 4". The second way saves time and avoids typos, but it helps to know what is actually going on when you let go of the mouse button.

The short answer

Dragging a file onto an assistant is a shortcut for "here is the material, read it". The assistant gets the file's contents, not your whole computer. It pulls the words (and sometimes the pictures) out of the file, puts them in front of the thing that is answering, and then your question is read together with that material.

That is all. There is no magic in the drop itself. What matters is what kind of file it was, how readable it was, and where the contents went afterwards.

What happens step by step

1. The file is identified. The program checks the type: a PDF, a Word-style document, an image, plain text, a spreadsheet. Each type needs a different reader.

2. The contents are extracted. For a typed document, the text is read straight out of the file. For a picture, a separate step has to look at the image to work out what it shows.

3. The result is trimmed to fit. Assistants can only consider so much material at once, so a long file may be cut, summarized or read in pieces. This limit varies from product to product.

4. Your question is attached. The extracted material and your question go together, so "question 4" has something to refer to.

5. The answer comes back, ideally pointing at where in the file it got the information.

File types and what they mean for reading

Not every file reads the same way. The big divide is between files that contain text and files that only contain a picture of text.

A quick test: open the PDF and try to select a single word with your cursor. If you can highlight it, the file has real text. If the whole page highlights as a block, it is a picture.

Walk-through: a worksheet PDF dropped in for an explanation

Back to the ramp worksheet. Say it is a typed PDF and you drop it in with the request "explain question 4 step by step".

1. The text of the page is extracted. Question 4 appears as text, along with the other questions.

2. The assistant finds the part you named. If you had said only "explain this" with nothing selected, it would have to guess which question you meant, so naming the number helps.

3. It reads the numbers: the cart gains 3 m/s every second, so the acceleration is 3 m/s². Time is 5 s, starting from rest.

4. It computes: v = a × t = 3 m/s² × 5 s = 15 m/s.

5. Check the units: m/s² times s leaves m/s, which is a speed. The answer fits the question.

Suppose instead the worksheet was a scan and the printed "3" smudged into something that looks like an "8". A recognizer might read "gains 8 m/s every second" and happily answer 40 m/s. This is the practical cost of scanned pages: the reasoning can be flawless and the answer still wrong because of one misread character. So with a scan, check the numbers the assistant quotes back against the paper.

Check what it read: Ask "what does question 4 say?" before asking for the explanation. If the quoted text is wrong, fix the source first.

What it can't do

• Read what is not in the file. A teacher's spoken hints, or a diagram that lives only on the board, are not in the PDF.

• Understand every picture. Charts and diagrams are harder than text, and small labels are often misread.

• Handle unlimited length. A 300 page textbook will not be considered all at once; you may need to point it at the chapter.

• Open locked files. Password protected or damaged files usually cannot be read at all.

Privacy and control

Think of dropping a file in as the same decision as attaching it to an email. Once the contents leave your computer to be processed, you should check the product's privacy page for what happens to them: how long they are kept and who can see them. If you do not know, assume the file is shared.

• Only drop files you are allowed to share. A friend's private notes, a school record or a company document may carry rules.

• Remove what is not needed. If page 2 holds a home address and you only need page 1, export page 1.

• Check the file for hidden extras such as comments or tracked changes, which can be read too.

Try it yourself

Pick a file you already have and test it. First try selecting a word to see whether it is typed or scanned. Then drop it in, ask the assistant to quote one sentence from page 2, and compare with the original. If the quote is right, you can trust the reading and move on to your real question.

DuckyHelper has homework files for this kind of job, and sees your screen when you press the key, so you can also just ask about the page that is open in front of you. You can try it for 10 minutes, no card. The same care applies to any tool: see data safety basics for AI tools at work for what to keep out of files you share.

Frequently asked questions

Is a screenshot the same as a file?

It is read like a photo of a page, so the words must be recognized from pixels. A sharp, straight screenshot works much better than a blurry one.

Why did it ignore the second half of my file?

Long files may be cut or summarized to fit. Point it at the section you need, or split the file.

Can it change my original file?

Reading and editing are different permissions. Dropping a file in normally only lets the assistant read a copy, but check each product to be sure.