Two modes sit in the toolbar.
Place image in Word (default) builds a real .docx with your picture fitted to the page. Use this when a form wants “a Word file” and you only have a scan.
OCR — extract text runs Tesseract in the tab and writes the letters as paragraphs you can edit. English is the default language pack; switch if the page is in Russian, German, Spanish, French or Simplified Chinese.
Getting better OCR
High-contrast scans of printed pages work. Photos of screens, handwriting and busy posters miss characters. Small type is upscaled before recognition.
The first OCR run may download language data from a CDN. After that, it stays cached in the browser. Place-image mode never needs that download.