Archives and libraries
An archive that's read, not just stored. Scanned books and records become text you can search, with their headings and tables. And when they can't leave the building, Baseer runs on your own servers or inside the Kingdom.
Baseer OCR
From a document image to ready data
Turns scanned pages, images and PDFs into clean Arabic text, with its headings, tables and reading order.
A closer look
A table stays a table.
Tables are extracted as HTML inside Markdown, so they stay as they are however complex. On the Misraj-DocOCR benchmark, Baseer scored the highest table structure accuracy (TEDS) among the systems the paper compared.
0.25
Word error rate on Misraj-DocOCR. Lower is more accurate.
Against 0.37 for Gemini 2.5 Pro. Source: Baseer paper, Table 5 (arXiv:2509.18174).
Headings and reading order
Headings, lists and columns come back in the order a person reads them. Because Baseer reads the whole page, not line by line.
Reads Arabic the way it's written.
Joined letters, diacritics, varied typefaces, right-to-left writing. This is where general tools stumble, which is why Baseer learned from 500,000 Arabic pages, real and synthetic.
Who it is for
For everyone who works with Arabic documents.
Publishers and education teams
From paper to text you can edit. Books, journals and teaching material become Markdown you edit directly. Page numbers, watermarks and images are separated from the text, so they don't mix into it.
AI and search teams
Arabic PDFs reach your search and training pipelines as structured text, not broken characters. It is how Seamless Enterprise makes scanned contracts searchable.
Developers
Send the page image, get Markdown back. Through Kawn Console, or as a step in Seamless API. To try it, upload a page at baseerocr.com.
Integrations
Works with the systems you have.
Works with the Misraj ecosystem.
Kawn Console System
Seamless API System
Seamless Enterprise System
Baseer Extract System
Cloud Deployment
On your servers Deployment
Inside the Kingdom Deployment
Get started
Upload a page. See for yourself.
Try Baseer on one of your own documents, or tell us about a whole archive.