Sovereign AI means an organisation keeps control over its data and over the AI models that process it: where they run, who can access them, and whether the data ever leaves its hands. Marsad is designed for this. It runs on open-weight AI models rather than a third-party chat service, isolates every workspace from the others, and can be deployed on infrastructure the customer controls, including fully offline.
Why sovereignty matters now
Generative AI has made it easy to paste a contract or a financial report into a public chatbot. For many organisations in Saudi Arabia and the Gulf, that is not acceptable. The data may be commercially sensitive, may include personal information, or may fall under sector rules on where and how it is processed. Saudi Arabia’s Personal Data Protection Law (PDPL), sector regulators, and internal security policies all push in the same direction: know where your data goes, and keep control of it.
The challenge is getting the benefits of AI (answers, summaries, monitoring, recommendations) without giving up that control.
Four questions to ask any AI platform
Which model processes my data, and who runs it?
Many AI products send your data to an external model provider’s API. Marsad uses open-weight models, whose weights can be downloaded and run by whoever operates the platform. The model can therefore run in an environment chosen by the operator, rather than as a call to someone else’s service.
Is my data used to train someone else’s model?
With a self-operated open-weight model, your documents and records are processed to answer your questions. They are not sent to train a third party’s model.
Is my workspace isolated from other customers?
In Marsad every workspace is isolated, and permission checks apply to every read, including the reads the AI performs when answering a question. The assistant can only see what the person asking is allowed to see.
Can it run on my own infrastructure?
Yes. Marsad can be deployed on infrastructure the customer controls, including an offline deployment with no dependency on external internet services. That suits organisations whose policies require data and processing to stay inside their own environment.
Sovereignty is also about decisions
Control over data is half the picture. The other half is control over what the AI does. In Marsad, AI proposes and people approve, and every decision is recorded in a tamper-evident ledger. You keep a verifiable record of who decided what and on what evidence. Read more in Can AI take action in your business safely?
Arabic-first, built locally
Sovereignty also means a platform built for the region. Marsad is built by NASL in Saudi Arabia. It is Arabic-first and works with Arabic documents, questions and answers, and it is designed around how organisations here actually work.
Frequently asked questions
What is the difference between sovereign AI and private AI?
The terms overlap. “Private AI” usually means your data is not shared with others. “Sovereign AI” goes further: control over where the models and data physically run, and under which jurisdiction.
Does sovereign AI mean lower quality answers?
Not necessarily. Today’s open-weight models are capable, and much of the answer quality comes from grounding: giving the model the right, permission-checked facts from your own data, and citing the source of each figure.
Can Marsad run without an internet connection?
Yes. An offline deployment option is available for environments that require it. Contact us to discuss your requirements.
Discuss a deployment that fits your policies: request a Marsad demo.