Arabic health records a system can actually read
Clinical Arabic, handled with the precision it requires
Healthcare organizations need AI that bridges everyday Arabic with the precise language of clinical records and diagnostics, deployable inside the same compliance boundary as any other regulated sector.
Why clinical Arabic needs a model of its own
Clinical language sits a long way from everyday Arabic, and a general embedding model treats the two as the same thing. Kawn-Embed-Medical is trained on clinical vocabulary, medical records and health diagnostics specifically to bridge that gap, and Mutarjim was built to preserve meaning across medical and technical content rather than to read fluently. Both deploy inside the same compliance boundary as any other regulated sector.
A model for the vocabulary
Kawn-Embed-Medical is a domain-specific Arabic embedding model for clinical vocabulary, medical records and health diagnostics. Not a general model asked to cope.
Patient data does not travel
On-premises and dedicated in-Kingdom deployment across the portfolio, so the capability can be used on material that is not permitted to leave.
Terminology that survives translation
Mutarjim was evaluated on demanding domains rather than on general prose, which is the difference between a translation that reads well and one that is correct.
What healthcare organizations deploy
- Clinical retrieval
Search that understands the vocabulary of the record.
Kawn-Embed-Medical converts Arabic clinical text into vectors tuned for that domain, so retrieval across records, protocols and internal guidance returns what a clinician meant rather than what the words happened to match. It runs as part of enterprise retrieval inside your own environment.
Enterprise Search & RAG - Private deployment
Capability on the inside of the boundary patient data sits behind.
The same private AI platform other regulated sectors deploy: strong open-source models, retrieval over your own knowledge, chat and analysis. Installed on-premises or on dedicated in-Kingdom infrastructure and architected around SDAIA and NCA ECC.
- Clinical translation
Medical meaning that survives the crossing between Arabic and English.
Mutarjim is a bidirectional Arabic-English model built on Kuwain and trained on a curated bilingual corpus for fine-grained alignment, with medical among the demanding domains it was built to preserve. It is small enough to run in production rather than only in a demonstration.
Translation & Localization
Deployment and compliance
On-premises and dedicated in-Kingdom deployment, inside the same compliance boundary any other regulated sector uses, architected around SDAIA and NCA ECC. The Trust Center sets out data residency and the deployment options in full.
Let's talk about what you're trying to build.
Tell us the problem. We'll tell you honestly whether AI is the right answer.