Skip to content
YallaConvert

Summarize a scanned document

Scans and photos have no text to copy. The summarizer reads them with OCR on your own device first, then summarizes what it found.

Summarize a scan
  • Needs a free account
  • Text is processed by Anthropic (Claude)
  • Encrypted in transit

What you get

  • OCR in your browser for scanned PDFs, up to 20 pages per document, and for JPG, PNG, WebP and HEIC photos.
  • The recognized text to look over before you summarize, so misread words are caught first.
  • A brief summary, key points, action items and important terms, written from the recognized text.
  • OCR is free and runs on your device. The summary itself costs credits: the cost is shown before you run it, and you are charged only if it succeeds.

Works best with

  • Printed pages that are flat, sharp and evenly lit. Blur, shadows and curled pages cause misread words.
  • Choose the document's language before OCR runs. It reads 14 languages, including Arabic, Chinese (Simplified), Hindi and Russian.
  • OCR here is built for printed text, so handwriting may not be recognized.
  • For scans longer than 20 pages, split the PDF and summarize it in parts.

How it works

  1. Drop the scan or photo into the AI Summarizer

    Sign in and add the PDF or image. For a scanned PDF the summarizer notices there is no text and offers OCR.

  2. Pick the language and read it with OCR

    Choose the document language and select Read the text with OCR. This step runs on your device and is free; only the summary uses credits.

  3. Review the text, then summarize

    Check the recognized text, try a sharper image if it is garbled, then run the summary.

Example

Illustrative example: a photographed letter, after OCR, summarized.

Input

Photo of a two-page letter from a landlord about a rent review

Output

  • Brief summary: The rent rises from 1 March; the letter explains the new amount and how to respond.
  • Action items: reply by the date given in the letter if you want to discuss the increase.

Frequently asked questions

Is the scan uploaded for OCR?
No. OCR runs in your browser with an engine loaded from our own site, and the image stays on your device. Only the recognized text is sent, encrypted, to Anthropic (Claude) to write the summary.
Which languages can OCR read?
English, Spanish, French, German, Italian, Portuguese, Dutch, Russian, Arabic, Chinese (Simplified), Japanese, Korean, Hindi and Turkish. Pick the one your document is written in before OCR runs.
What if the OCR text has mistakes?
Look over the recognized text before you summarize. If it is garbled, try a sharper photo, a flatter scan or a different document language, and run OCR again. Re-running OCR is free; credits are charged only when you summarize.

Related

Summarize a PowerPointKey points and action items from a .pptx deck.Explain a PDF in plain EnglishDense documents explained, with deadlines and costs flagged.Translate a PDFThe text of a PDF in one of 40 languages.AI SummarizerSummarize a PDF, Word or PowerPoint file, an HTML file, a photo or pasted text: key points and action items.
Open the AI Summarizer