The model runs on your machines
The AI model is downloaded once during setup — from your own server or the internet — and from then on runs locally. There is no call to an outside service to generate an answer.
Privacy
Sanad runs on employee devices or your organization server. Documents, conversations, and generated work stay within that controlled environment.
The AI model is downloaded once during setup — from your own server or the internet — and from then on runs locally. There is no call to an outside service to generate an answer.
Files you add are read and indexed on employee devices or your organization server. They are not sent to Sanad's vendor services or used to train a vendor model.
After models and licensing are provisioned, core work can continue with the internet unplugged, subject to the applicable entitlement or roaming-grace window. You can verify this during a controlled demo.
Licensing and update services receive the identifiers, versions, and dates needed to issue entitlements and deliver releases. They do not receive prompts, documents, answers, transcripts, notes, or generated work.
How activation works, exactly what we log, the cryptographic model, and our compliance posture are shared with your IT and procurement teams during evaluation — reviewed against your own requirements rather than published as marketing.
Email hello@sanadhq.app and we'll send the relevant documents and arrange a call with whoever needs to sign off.