Sovereign AI platform
Hekima
The power of generative AI, without your data ever leaving the building.
Hekima, “wisdom” in Swahili, is a private generative-AI platform: a secure AI assistant, answers grounded in your own documents and AI agents under strict control. It runs on your infrastructure, in your private cloud, or even completely offline.
Platform operational · pilot deployments open
The problem
Your data can’t go to a public AI cloud
Confidentiality and data-residency rules rule out most AI services, and the ones that remain leave gaps in your audit trail.
Most AI tools assume you are always online
Secure sites, isolated networks and critical operations need AI that keeps working with no connection to the outside world.
Automation without accountability
AI agents that act without approval, limits or logs are a risk no regulator will accept under the EU AI Act, DORA or NIS2.
Who it is for · Built for organisations whose data cannot leave controlled infrastructure: public sector, finance, insurance, healthcare, energy and telecom, and for the IT service providers who host AI for their own clients.
What it does
A private AI assistant
Chat that automatically picks the right model for each question: fast for simple ones, more powerful for complex ones.
Answers from your own documents
Ask questions of your policies, contracts and procedures, and get answers that cite their sources.
Many organisations, strictly separated
Each client or department gets its own sealed space. Service providers can offer it under their own brand.
Enterprise-grade access control
Single sign-on, multi-factor authentication, roles, quotas and usage analytics, ready for your security review.
Deployed where you decide
Your data centre, your private cloud or a fully air-gapped site, with licences that work offline.
No lock-in to one AI vendor
Hekima works with several AI models and switches between them, so you keep your options open.
LLM engineering with governance built in
Hekima is where we apply what we advise our clients: AI that stays accountable, inspectable and enforceable.
- Every question, answer and administrative action is recorded in a complete audit trail.
- Personal data is kept out of monitoring data, and uploaded files are scanned before use.
- Sensitive automated actions wait for a human approval, following rules you define.
- Compliance dashboards and reports give your risk and audit teams what they need.
Where we are
The core platform is running and has been through several rounds of security hardening. Governed AI agents and offline knowledge modules are next on the roadmap. We are opening a limited number of pilot deployments.
Join the project
We are looking for people who want to help shape what comes next.
Builders
Engineers, data scientists and designers who want to work on applied AI that has to be right, not just impressive.
Investors
Partners who see the opportunity and want to help the product reach its market faster.
Pilot clients
Organisations ready to try it on a real case, with our team at their side and a direct say in the roadmap.
We keep the inner workings confidential. Demos, technical details and figures are shared in person, under a non-disclosure agreement.
Start the conversation