what is docslight fr?

DocSlight is an open-source tool for pulling readable text and structured data out of documents. It handles PDFs, scanned pages, plain images, and Office files like Word, PowerPoint, and Excel, converting them into clean Markdown or JSON that other software can use. It combines optical character recognition covering more than eighty languages with AI-based parsing that tries to identify structured fields, such as the line items on an invoice, rather than just dumping raw text. Note that this specific repository's README says it has moved to a new location on GitHub, so it may no longer receive updates here. The tool can run in two ways. Local mode installs through pip and processes documents entirely on your own machine at no cost and without an account, though it needs a locally running AI model for the deeper structured extraction features. Cloud mode sends documents to the maker's hosted service using an API key, which trades some privacy for higher accuracy and does not require you to run your own AI model. There is also a browser based web interface you can run yourself with Docker, where you drag and drop files to parse them, plus a command line tool and a Python library for developers who want to call it from code. The README points to specific uses such as feeding documents into retrieval systems for AI chatbots, processing invoices, analyzing contracts, and giving AI agents a way to read documents as a tool. A paid enterprise edition adds things like workflow automation, access controls, audit logs, and support for private company deployments. This project fits developers and companies building document heavy AI applications, from personal projects to enterprise systems, rather than someone looking for a simple one-off file converter. The full README is longer than what was shown.