Compliance, built in
GDPR, NIS2, and the EU AI Act keep raising the bar. When the model runs on hardware you own, the question “where is our data processed?” has a one-word answer: here.
Sovereign on-premises AI · Slovenia & DACH
Run intelligence on your terms.
runint puts modern AI inside your company — on servers you own, in your building, on your network. Your contracts, records, and know-how stay exactly where they are. Nothing ever touches a public cloud.
Why on-prem
AI that never leaves the building
Your IP stays yours
Contracts, patient records, designs, financials — the things you would never email to a stranger should not be sent to a foreign cloud API either.
Cloud-class capability
Open-weight models like Llama and Mistral now handle everyday knowledge work — drafting, summarising, extracting, searching. Run locally, they are fast, private, and yours.
The offering
A managed AI appliance
One turnkey service: we spec it, install it, and keep it running. You get working AI on your premises without hiring an ML team.
Hardware, right-sized
From a high-memory workstation to a single GPU rack — we spec a machine that fits your workload, and you buy it directly from the distributor. No markup, no lock-in: the hardware is yours.
Models & engine
Quantised open-weight models (Llama, Mistral) served by production engines such as vLLM and Ollama — tuned to your hardware and able to run fully air-gapped.
Secure document search
Enterprise retrieval (RAG) over your internal documents: ask questions in plain language, get answers with sources — without a single file leaving your network.
Operations, handled
We monitor system health, apply security patches, and upgrade model weights as better open models are released. You use AI; we keep it running.
Process
From first call to running system
-
Assess
We map your use cases, data, and compliance requirements, then produce a precise hardware and software specification.
-
Procure
You purchase the hardware directly from the distributor, to our verified spec. You own it outright from day one.
-
Deploy
We install the machine on site, integrate it with your network and data, index your documents, and hand over a working system.
-
Operate
A monthly Managed AI SLA covers orchestration, monitoring, security patches, and model upgrades — with an engineer you actually know.
Pricing
Transparent by design
Hardware
At cost
Purchased by you, directly from distributors, to our verified spec. Your asset, on your balance sheet.
Onboarding
One-time fee
Physical installation, model setup and quantisation, network integration, and initial document indexing.
Managed AI SLA
Monthly
Ongoing orchestration, health monitoring, security patches, and model weight upgrades.
Every engagement is scoped individually — ask for a quote.
Who it’s for
Built for the European mid-market
Firms of 50–500 people: big enough that AI moves the needle, regulated enough that the public cloud is a liability. We work across Slovenia, Germany, Austria, and Switzerland.
- Legal
- Healthcare
- Accounting
- Manufacturing
Contact
Talk to an engineer, not a sales deck
runint is run by an AI-infrastructure engineer working in Germany’s software industry. The person who specs your rack is the person who answers your email.