An AI server for your company, with open models and a search engine over your own documents. Your team uses it from the browser, the data never leaves your network and you pay nothing per token. You pay for the hardware, outright or on renting, and the set-up once: no monthly fees.
Your team asks it like a colleague; it answers from your own documents and shows where each answer comes from.
Ask in plain language and it answers with the exact document and paragraph: contracts, case files, reports or internal rules.
Summaries of long documents and drafts of letters, emails or reports based on your own templates.
You pay for the hardware and the set-up once. Using it more costs nothing extra, and there are no message caps per person.
Adding someone costs nothing: the hardware is chosen once for your headcount and there is no monthly fee.
With a cloud AI, every document you give it leaves your company. With private AI, everything is processed on your server.
Documents and questions are processed on your server. They do not travel to third-party servers or get used to train anyone’s models.
Contracts, medical records, payroll, drawings or bids: you can work with them because you never send them to an outside provider.
No international transfers and no extra AI provider in your processing records: the processing stays on your own systems.
A username and password for each person, with access only to the document collections you assign.
It runs without the internet. It only needs a connection to download security updates, and you decide when it connects.
The model does not change unless you decide so, and you do not depend on price rises, new terms or outages of a cloud service.
Does your company work under ISO 27001 or ENS? Read why many companies restrict ChatGPT with confidential data and how private AI fits.
The hardware is sized by headcount, and you can buy it or rent it: from €2,903 for up to 5 people to €12,413 for up to 60. The tiers are below the calculator.
Set-up is done in our workshop and remotely; if you need a technician to come to your premises, travel is invoiced separately. Integrations with other software and later changes are quoted separately.
Both have their place. This is what really changes between them.
| Private AI | Cloud AI | |
|---|---|---|
| Where your documents are processed | On your server, inside your network | On the provider’s servers |
| Confidential data | Stays in the company | Leaves the company with every query |
| How you pay | Hardware and set-up, once | Per person or per token, every month |
| If you use it more | Costs the same | Costs more |
| Without an internet connection | Keeps working | Does not work |
| The largest models on the market | No: open models | Yes |
| Up-front investment | Yes: hardware and set-up, or the hardware on renting | No |
Enter how many people would use it and what cloud AI costs per person. You will see what you would pay the cloud every month, what your own server costs with the set-up included and how many months it takes to pay for itself.
Hardware for up to 15 people: Ryzen 9 9900X · 64 GB · RTX 5070 Ti 16 GB · €4,727
With 10 people at €100 a month, cloud AI would cost €36,000 over 3 years; private AI, €5,497, and the hardware is yours. It pays for itself in 6 months.
| When | Private AI | Cloud AI | Saving |
|---|---|---|---|
| Month 12 | €5,497 | €12,000 | €6,503 |
| Month 24 | €5,497 | €24,000 | €18,503 |
| Month 36 | €5,497 | €36,000 | €30,503 |
Published prices per person per month, excl. VAT, October 2026, at $1.14/€: business plans such as ChatGPT Business, Claude Team or Microsoft 365 Copilot, $20-30; advanced plans such as ChatGPT Business Premium or Claude Team Premium, $100-125; top plans such as ChatGPT Pro or Claude Max 20x, $200. Hardware by headcount, at configurator prices: up to 5, €2,903; up to 15, €4,727; up to 30, €6,509; up to 60, €12,413 (details below). Set-up and configuration: €770, one-off and with no monthly fees. On renting, indicative 36-month hardware fee (3.20% of the price per month). Prices excl. VAT.
I want this proposalRyzen 7 9700X, 32 GB of RAM and a 2 TB SSD. Models up to about 20B, such as gpt-oss-20b.
€2,903or about €93/mo on rentingRyzen 9 9900X, 64 GB of RAM and a 2 TB SSD. The same models, with more speed for more people at once.
€4,727or about €151/mo on rentingRyzen 9 9950X, 64 GB of RAM and a 4 TB SSD. Two graphics cards: models of about 30B and longer documents.
€6,509or about €208/mo on rentingRyzen 9 9950X, 128 GB of RAM and a 4 TB SSD. Professional graphics cards: models up to about 70B, or two models at once.
€12,413or about €397/mo on rentingPrices from the configurator, excl. VAT and with a 3-year warranty. If you need a different configuration, we adjust it with you.
Open models improve every few months and are not paid per person. When a better one comes out, it can be installed on the same server; and if one day you need more, you add memory or a graphics card instead of replacing the machine.
Neither the documents nor the questions leave your network. There is no outside provider that can read them, keep them or use them to train its models.
For ISO 27001 or Spain’s ENS it is one more internal system, not a cloud AI supplier to assess and monitor. Read why.
No per-person fee and no provider price rises: adding someone or using it more costs nothing extra.
Contracts, case files and filings: find the precedent and prepare the draft in minutes.
Protocols, reports and internal documentation, consulted without data leaving the centre.
Standards, specifications and project documentation, with answers that cite their source.
Tell us how many people you are and which documents you work with. We send the proposal in under 12 business hours.