Book a consultationPrivacy by design · your data stays in your own Odoo · no forced US cloudSovereign AI from €690 →
MindDesk
MindDesk Sovereign Box

Your own AI server.
On your premises.

A preconfigured, managed AI server sitting alongside your Odoo. Leading open models run locally - by default, prompts and documents stay on your server. We deliver, set up, and maintain it. You keep full control.

Set up by usLocal models by defaultRemotely managed from GermanyNo US cloud contract
Leading open models included Privacy by design Atomic updates with rollback Advance replacement on failure
Sovereign Box · Status
OnlineAppliance reachable, local models loaded
LocalModels answering on your box
128Requests today - all processed locally
Local
models instead of cloud, by default
< 1 day
set up and looked after by us
24/7
monitoring & remote management
every 3
months, a model update
The decisive difference

Most AI tools send your data to the model. We send the model to your data.

On the Sovereign Box, all the computing happens on-premises: chat, answers, and knowledge processing on local models, your knowledge stays in your own database. MindDesk sits above it as a control layer - with approvals and a complete audit trail. The result: local models instead of cloud, no US cloud, no surprise on the invoice.

Local - models run on your own server by default
Managed - we maintain it remotely
Fair - buy once instead of endless cloud fees
sovereign-box · documents
Summarize the contract with Müller Bau.
local models by defaultPayment term 30 days, special terms from €5,000 order value. Processed on your box, using the local models.
File the summary as a document for me.
I'll file the contract summary in your documents.
File itNot needed
How it works

Three layers. All under your roof.

No detour via third-party servers by default - the models come to your data.

1

Local models on your own server

Chat, answers, and knowledge processing all run on the box.

Compute stays on-premises

Every request is answered by the local models on your box. Prompts and documents stay on your server by default - local models instead of cloud is the default.

Leading open models

Your box runs current, high-performance open AI models - pre-selected and configured to match the size of your organization.

No subscription meter ticking away

No token clock running per request. Your team uses the AI as much as it wants - the compute is yours.

2

Your knowledge stays in your own database

Documents, context, and history - all on your own infrastructure.

Your data, your location

Your processed company knowledge sits in your own database alongside Odoo - not in a third-party cloud. You keep full ownership of it at all times.

Answers with a source

Whatever the AI answers comes from your own documents - with a reference to where it found it. Traceable, not guessed.

Delete means delete

Because everything lives with you, you decide on retention and deletion. No data residue on third-party servers.

3

MindDesk as the control layer above it

Approvals and an audit trail over everything the AI does.

Approvals you set

You decide what the AI can do on its own and what needs your go-ahead. Sensitive actions always stay gated behind an approval.

Complete audit trail

Every AI action is recorded and traceable, and field changes can be undone - the foundation for audit-ready EU AI Act evidence.

Per-user, with permissions

The AI only sees what that particular employee is allowed to see. Your Odoo permissions apply - no detour, no back door.

What's inside the box

Preconfigured. Managed. Ready to go.

You get a finished appliance, not a build-it-yourself project. Set up, remotely maintained, and legally clean - you plug it in, we take care of the rest.

Leading open models included

The right AI models are selected, installed, and tuned to work together - you don't have to procure, license, or configure anything.

everything set up

Atomic updates with rollback

Updates are applied in a single step - if anything goes wrong, the box automatically reverts to the last working state. No maintenance risk for you.

remotely managed

Compact GPU appliance

A quiet, energy-efficient device for your office or server room - depending on tier, from the 128 GB mini appliance up to the workstation GPU. No data center required.

plug & ready

privacy by design, no US cloud contract

No contracts with US providers, no data transfer across the Atlantic. Processing stays on your premises by default - evidence to support your GDPR and EU AI Act assessment.

sovereign
Bottom line: an AI server you don't have to run yourself - delivered, set up, and remotely managed, yet entirely on your premises.
Find the right box →
Choose your size

A box for every organization

From the compact office device to the workstation GPU for large enterprises. Every box ships preconfigured and with Managed Sovereign Service.

Sovereign S - Office
5-25 employees
Today's quotehardware, setup and monthly care in one offer
  • 128 GB mini appliance
  • leading open models included
  • up to ~10 AI users
  • local models instead of cloud, local
Popular
Sovereign M - Team
10-50 employees
Today's quotehardware, setup and monthly care in one offer
  • GPU appliance, 128 GB
  • strong knowledge processing
  • 10-25 concurrent users
  • quarterly model updates
Sovereign L - Enterprise
50-250 employees
Today's quotehardware, setup and monthly care in one offer
  • Workstation GPU
  • multi-company
  • 25-100+ concurrent users
  • priority remote support

More than 250 employees, or special requirements? We plan larger setups individually on request.

Match the right size →
Managed Sovereign Service

We keep the box running in top shape - remotely

The ongoing service is what separates a server you have to babysit from one you never have to think about.

24/7 monitoring

We keep an eye on your box's health, load, and security around the clock - and act before you ever notice a problem.

Security updates with rollback

Security updates arrive automatically and atomically. If anything doesn't run cleanly, the box reverts itself to the last good state.

Quarterly model updates

Every three months we bring the current best open model onto your box - without you having to do anything. Your box never ages.

Advance replacement on failure

Should hardware fail, you receive a replacement before you send back the faulty unit. Your operations never grind to a halt.

Monthly report

Once a month, a clear report: usage, health, updates. You always see exactly what your box is delivering.

Support from Germany

A human who knows your setup - reachable, easy to talk to, no time-zone detour. Support delivered directly from Germany.

A month with the box

Day-to-day stays invisible - until you ask

Updates run automatically in the background, the models stay on your box. On request, you get the status in one sentence - including the date for the next model update.

sovereign-box · monthly report
How did the box's month go?
Monthly report1,842 requests answered entirely locally, all security updates applied automatically.
And the next model update?
Scheduled in 9 days, runs automatically with rollback protection.
Apply nowKeep as scheduled
The promise: your box always runs the current best open model - secure, current, without you lifting a finger.
Learn about the service →
Delivery & go-live

Plug it in. Quick setup. Get going.

No on-site project, no weeks-long rollout. The box arrives fully configured - the rest is done in a few hours.

  • 1 Delivered preconfigured - matched to your Odoo and your size.
  • 2 Plug in power and network - that's the extent of the effort.
  • 3 1-2 hours of remote onboarding with us - access, first use cases.
  • 4 Remote onboarding - no on-site visit needed.
sovereign-box · status
Box plugged in - what now?
MindDesk · Onboarding Box detected, local models ready. Remote onboarding started: Odoo connected, approvals set, two use cases active.
LiveView report
Delivered · set up · remotely managed

Your AI server, without you having to run it

We match the right box to your size, deliver it ready to go, and keep it in top shape for good with the Managed Service. You keep full control.

Find the right box Compare tiers
Frequently asked questions

What customers ask about the box

Where do prompts and documents go?+
Chat, answers, and knowledge processing run on your box's local models, your knowledge stays in your own database. Prompts and documents stay on your server by default - local models instead of cloud is the default.
Do I have to maintain or update the AI models myself?+
No. The Managed Sovereign Service handles monitoring, security updates with automatic rollback, and quarterly model updates. As a result, your box always runs the current best open model - without you lifting a finger.
How long does setup take?+
The box arrives preconfigured. You plug in power and network, we handle a 1-2 hour remote onboarding - live in under a working day. No on-site projects, no weeks-long rollout project.
What happens if the hardware fails?+
The Managed Service includes advance replacement: you receive replacement hardware before you send back the faulty unit. That way your operations don't stall while it's being repaired.
Do I need my own data center for this?+
No. The box is a compact, quiet device for your office or server room - power and network are enough. For larger setups there's the workstation GPU variant; above 250 employees, we plan individually on request.
What does the box and the service cost?+
Server hardware prices change every few weeks, so we quote on the day you ask: hardware, setup and the monthly managed service in one offer. Larger tiers (M, L) scale hardware, concurrent users, and service accordingly. You buy once instead of paying endless cloud fees.
Preconfigured · local models instead of cloud · set up and looked after by us

Get your own AI server on your premises

Sovereign AI on your hardware, under your rules - delivered, set up, and remotely managed for good. Let's match the right box to your organization.

Find the right box View pricing