Document OCR to DOCX converts your uploaded txt, csv, docx, xlsx, pptx, odt, ods, odp, or pdf documents into one editable Word document (named YYYYMMDDHHMMSS.docx). It runs as a multi-round AI Agent workflow: each new conversation creates a notebook workspace named YYYYMMDDHHMMSS in the background, the system first renders every page of the document into a page image, and the site's configured vision model then performs optical character recognition on each page in order. Pictures embedded in the document are cropped into separate image files and inserted into the Word file as embedded pictures, each with a caption made from the text found inside that figure; tables are re-created as real Word tables based on their content; heading levels and reading order are preserved so the Word layout stays as close to the source document as possible. When all pages are assembled in order, the .docx file is generated with a download link, and each cropped figure gets its own download link as well. Usage is simple: upload one or more documents and press send. The conversion streams each round's progress and tool actions in real time, can be interrupted at any time with the stop button, and you can continue the same conversation to request adjustments. The document content keeps the source document's original language — it is a conversion, not a translation. You can also open this conversation's notebook on the Notebook page to review the page images and generated files in its workspace. Document OCR to DOCX runs many rounds of actions in a row and calls the vision model page by page, using a large amount of tokens; at least 100,000 credits are required in your account to use this feature. Maximum total upload size per request: 256MB (all files and message content combined). Output content formats: Text Accepted document formats: txt, csv, docx, xlsx, pptx, odt, ods, odp, pdf. File output: an editable Word document named YYYYMMDDHHMMSS.docx (embedded pictures are cropped and inserted with captions; tables are laid out as Word tables).