Services
Private AI — Hosted & On-Prem
In short
Private AI means your assistant runs on a model that you control instead of a public AI service. With Private Hosted, it runs on a dedicated GPU we manage in the EU; with On-Prem, it runs on hardware inside your office. Either way, your documents and customer data are never sent to OpenAI, Google or Anthropic.
Who needs private AI?
If your AI will read patient files, contracts, case notes, payroll or financial statements, sending that data to a public AI API may not be acceptable to you, your clients or your regulator. Private AI gives you the same capabilities — chat, document Q&A, summaries, translation — without the data leaving your control.
Private Hosted: a dedicated model on a dedicated GPU
We run an open-weight model (tuned for Arabic and English) on a GPU server dedicated to you, in an EU data centre. Nobody else's data is on it, and no AI company sees your prompts. We handle updates, monitoring and backups. You get a data processing agreement.
On-Prem: nothing leaves the building
We source and install a workstation or small server in your office, connect it to your document folders, and train your team. It works even without internet. Ideal for law firms, accountants and medical groups with strict confidentiality.
What can it do?
- Answer questions about your contracts, policies or procedures — with page citations.
- Summarise long documents and highlight obligations, deadlines and risks.
- Translate Arabic ↔ English documents while keeping legal terminology.
- Power the same WhatsApp or website agent, privately.
Is a private model as good as ChatGPT?
For focused business tasks on your own documents, modern open-weight models are very capable, especially with good retrieval. Try it yourself in the Private document assistant demo — you can switch between cloud and private and compare.
Cloud AI vs Private Hosted vs On-Prem
Cloud AI Runs on leading AI providers (EU/US) | Private AI — Hosted A dedicated private model on a dedicated GPU we manage | Private AI — On-Prem On hardware in your office. Nothing leaves the building | |
|---|---|---|---|
| Where data goes | Processed by the AI provider, not used for training | Stays on our dedicated server. Never sent to OpenAI / Google / Anthropic | Never leaves your office network |
| Setup time | ~7 days | ~2 weeks | 3–5 weeks (incl. hardware) |
| Monthly cost | from €79 | from €490 | from €290 maintenance |
| Best for | Most SMBs: salons, restaurants, car rentals | Clinics, agencies, anyone with GDPR concerns | Law firms, accountants, medical groups |
Cloud AI
Runs on leading AI providers (EU/US)
- Where data goes
- Processed by the AI provider, not used for training
- Setup time
- ~7 days
- Monthly cost
- from €79
- Best for
- Most SMBs: salons, restaurants, car rentals
Private AI — Hosted
A dedicated private model on a dedicated GPU we manage
- Where data goes
- Stays on our dedicated server. Never sent to OpenAI / Google / Anthropic
- Setup time
- ~2 weeks
- Monthly cost
- from €490
- Best for
- Clinics, agencies, anyone with GDPR concerns
Private AI — On-Prem
On hardware in your office. Nothing leaves the building
- Where data goes
- Never leaves your office network
- Setup time
- 3–5 weeks (incl. hardware)
- Monthly cost
- from €290 maintenance
- Best for
- Law firms, accountants, medical groups
What it does for you
Law firms
Ask questions across contracts and case files; draft summaries; translate Arabic ↔ English — without data leaving the firm.
Clinics
Internal assistant for policies and procedures; patient-facing agent without sending health data to US AI companies.
Accountants
Extract figures from statements and invoices, answer staff questions about client files, all on local hardware.
How much does it cost?
Private AI — Hosted
The same agent, or a document assistant, running on a dedicated private model on a dedicated GPU we manage. Your data is never sent to OpenAI, Google or Anthropic.
Setup from €1,490
+ €490 /mo from
- Dedicated GPU (EU data centre)
- Open-weight model tuned for Arabic + English
- Document Q&A with page citations
- GDPR-friendly: DPA, EU hosting, no training on your data
- Monitoring & updates included
Private AI — On-Prem
Local AI installed on hardware in your office, connected to your contracts, policies or patient files. Nothing leaves the building. Hardware sourcing included.
Project from €4,900
+ €290 /mo maintenance from
Hardware cost depends on team size; quoted separately at cost.
- Hardware sourcing & installation
- Works offline, inside your network
- Connected to your document folders
- Staff training session
- Remote maintenance & model updates
FAQ
Is private AI GDPR compliant?
Private AI makes GDPR compliance much simpler: data stays in the EU (hosted) or in your office (on-prem), there is no third-party AI processor, and we sign a DPA. Compliance also depends on your own processes, which we help you document.
Which models do you use?
Current open-weight models with strong Arabic and English (for example Qwen, Llama or Mistral families), chosen and tested for your use case. We can switch models as better ones are released.
What hardware do I need for on-prem?
For a small office, a single workstation with a modern GPU is usually enough. We size it to your team and document volume and quote hardware at cost.
Can I start in the cloud and move to private later?
Yes. The agent's knowledge and configuration move with you.
See it answer your customers before you pay anything.