Sovereign alternative to ChatGPT Enterprise, Microsoft Copilot, Azure OpenAI
Sovereign Private GPT: your team's internal ChatGPT, with RAG
A centralized sovereign AI, operated for you in Europe. A ChatGPT-style chat interface with RAG over your documents, routed to your local models or a sovereign cloud AI. Unlimited users, you stay in control.
What the managed service covers
Private GPT combines Open WebUI (chat and RAG) with our AI gateway, which directs each request to your local models or a sovereign cloud AI. Bunker hosts and operates the whole stack in Europe, where your documents and their vectors stay stored.
Deployment and updates
We install and configure the software, then deploy the vendor's corrective releases after checking their stability. You have no server to administer.
Encrypted backups
Daily encrypted backups, kept in Europe. A restore is one ticket away.
Monitoring and security
Availability monitoring, security patches and network filtering, under the shared responsibility set out in our General Terms and Conditions of Service.
Human support
A team that answers on software it operates itself.
- Chat + RAG over your internal documents
- Routing to your local models or Mistral La Plateforme (EU)
- Disallowed personal data blocked before the cloud
- Hosted in Europe, unlimited users
Hosted in Europe, outside US jurisdiction
Without an internal tool, teams paste company data into consumer AI assistants (Shadow AI) and the business no longer controls where its content goes. A centralized AI hosted in Europe brings those uses back under your governance. Documents and their vectors stay stored in Europe, and disallowed personal data blocks the cloud call rather than leaking.
- Servers physically in Europe, operated by a European company.
- A European company, outside the scope of the US CLOUD Act and FISA.
- No subcontracting to a US hyperscaler.
- Re-internalizable: you can bring the service back onto your own servers.
Frequently asked questions
Where do my data and documents go?
Chat, documents uploaded for RAG and their embeddings, computed locally, are hosted in Europe by Bunker. On the general route, the retrieved excerpts go to Mistral with the question, and disallowed personal data blocks the request (fail-closed). To keep every excerpt in-house, route those uses through the sensitive route, on the private backend.
Which models does our AI gateway route to?
Our AI gateway directs each request to your local models (a private backend you provide) or Mistral La Plateforme (sovereign EU cloud), on your own account. No US provider is used by default. Routing is done by intent and by reasoning effort, and disallowed personal data blocks the cloud call (fail-closed).
What is included in the plan?
Managed sovereign hosting of both services, support, updates, backups and monitoring, from €190 excl. VAT/mo for unlimited users. The IT & AI diagnostic, leadership enablement (executive team and IT department) and training are offered as a quote, outside the plan. You pay your Mistral tokens directly, on your own account.
How is data control guaranteed?
Private GPT only delivers on its promise if public consumer AIs are blocked on the client network, otherwise teams route around the tool. Bunker provides the list of domains to block and change-management support, but does not operate that network blocking. See the CISO and IT recommendations in the documentation.
Deploy Private GPT in Europe
We handle hosting, updates and backups. You keep control of your data, and the ability to bring everything back in-house whenever you decide.