The useful result
Turn a scanned file into usable text, verify the key numbers and ask page-referenced questions.
- Turn a scanned file into usable text, verify the key numbers and ask page-referenced questions.
You upload a PDF and the assistant says it cannot find any text, or it answers with nonsense. Often the file is a scan: a picture of a page rather than words. A person can read it; software that expects real text cannot, until it is converted by optical character recognition, or OCR.
Google Drive can convert a PDF or image to editable text by opening it with Google Docs, and its help page lists the limits: a file of 2 MB or smaller, upright pages and text at least 10 pixels high. Other PDF and scanning apps offer similar “recognise text” tools. This lesson shows the test, the conversion and the checks that stop OCR mistakes becoming your mistakes.
The examples below are illustrative exercises, not recorded results from any tool. Sources were checked on 24 October 2026; menus, limits and prices change, so confirm details on the linked pages.
Before you start
- A PDF or photo of a document that the AI cannot read
- Google Drive or another tool with a text recognition feature
- The original file kept untouched
Review evidence
Documentation-based review · 2026-10-24
Primary documentation and official guidance checked on 24 October 2026. Examples, checks and exercises are original illustrations, not recorded tool tests. Cover artwork is a conceptual illustration.
Test whether the PDF has real text
Open the file and try to select a sentence with the cursor, or use the search shortcut to look for a word you can see. If nothing can be selected or found, the file is an image of text. Rotate it upright and make sure the pages are sharp with even lighting.
Check your result
You know whether the text is selectable, and the pages are upright and legible.
Convert the image to text with OCR
In Google Drive, upload the file, right-click it and choose Open with, then Google Docs. Check the limits on the help page: 2 MB or smaller, text at least 10 pixels high. For bigger files, use the text recognition feature in your scanner or PDF app. Lists, tables and columns often do not convert well, so expect to fix them.
Check your result
You have an editable text version, and you have noted which tables or lists need manual repair.
Check key numbers against the page, then ask by section
Compare names, dates and every important number in the text with the original image, because OCR confuses characters such as 0 and O or 1 and l. Then paste one section at a time and ask questions that require the page number and a quote. Work through long documents in parts.
Answer using only the text below, which comes from pages [range]. For each answer give the page number and the exact quote. If the text seems garbled or incomplete, say so instead of guessing. Question: [your question]
Text:
[paste section]Example / expected result
Illustrative example; not a recorded model result.
The OCR text shows “Total 1,2O0” while the image shows “1,200”. Correct before asking for any calculation.
Check your result
The numbers you will rely on match the original page, and answers include page references and quotes.
When it goes wrong
OCR struggles with handwriting, small print, stamps and tables, so check those by eye. Do not upload confidential documents to a service you have not checked. Help-page limits change, so confirm the current ones.
Make it your own
Scan one page of a printed document, convert it with OCR and find at least two errors by comparing with the picture. Write down which characters were confused.
Useful next resources
Sources and review
Last checked Oct 24, 2026. Features and account limits can change. Verify important results against your original material.
Did this help you complete your task?
Tell us what needs improving. No account or private output required.
Report incorrect information or a broken linkGenius Forges Editorial
Practical AI lessons with checked sources, labelled examples and human verification.
