1. Home
  2. Glossary
  3. Optical character recognition (OCR)
AI glossary · Media & modalities

Optical character recognition (OCR)

Optical character recognition (OCR): Optical character recognition (OCR) is technology that converts text in images, scans, and PDFs into machine-readable, searchable, editable text. Modern AI models extend it to read messy layouts, tables, forms, and handwriting.

Traditional OCR engines, such as the open-source Tesseract or the OCR built into Adobe Acrobat, recognize printed characters and return text. They work well on clean scans and poorly on skewed photos, faint receipts, or handwriting. Newer document AI services from Microsoft, Google, and Amazon add layout understanding, pulling out tables, key-value pairs, and fields from invoices and forms.

Multimodal language models have changed the everyday experience. You can now drop a photo of a whiteboard, a screenshot of a table, or a scanned contract into ChatGPT, Claude, or Gemini and ask for the text, a structured summary, or a spreadsheet. The model reads the image and the content in one step, and handles odd layouts and mixed handwriting better than classic OCR.

Accuracy is the catch. OCR errors are silent: a 6 becomes an 8, a decimal moves, a line is skipped. For anything with numbers that matter, such as invoices, bank statements, or medical records, verify totals against the original and spot-check entries. Also remember that a scanned document can contain PII or PHI, so run it only through tools your organization has approved for that data.

Example at work

An accounts payable clerk photographs a stack of 40 vendor receipts and asks an assistant to produce a spreadsheet with vendor, date, amount, and category. It takes a few minutes instead of an hour of typing. She then reconciles the total against the expense report, catches two amounts the model misread, and fixes them.

Why it matters

A huge share of business information is trapped in scans, photos, and PDFs. OCR, especially with modern multimodal models, lets you search it, analyze it, and automate around it. Verifying the numbers is what keeps the time savings from turning into an accounting error.

Related terms