A local LLM is a language model that runs on your own computer or server, so company data never leaves the company. I deploy open-source models with Ollama for companies in Slovakia and Czechia that need on-premise AI without the cloud.
I deploy a local model when data must not leave the company or when it pays off. Not because it sounds safer.
Who it is for
- Companies with sensitive data: healthcare, insurance, law, manufacturing
- Companies that can't send data to third parties because of GDPR or contracts
- Teams that want predictable costs instead of paying per token
If you need the strongest model for hard reasoning and the data isn't sensitive, a cloud API will be cheaper and better.
How it works
- We map the tasks and data and size the model to your hardware.
- I test several models on your tasks and compare accuracy and speed.
- I deploy the model with an API, logging and access rights.
- I set up quality measurement and a process for swapping models.
What you get
Model comparison
Several open-source models tested on your tasks: accuracy, speed and memory needs.
Hardware recommendation
A specific setup based on model size and number of users.
Deployed model with an API
The model runs on your server, with logging and access rights.
Model replacement procedure
When a better model comes out, you test it on the same set and swap it without a rebuild.
In practice
Construction Graph RAG runs entirely on one computer with a 4B model. ClaimPilot processes insurance claims with local models only: the larger Qwen3 30B-A3B matched the accuracy of a 14B model and was 2.4 times faster.
- Construction Graph RAG
99.2 % accurate answers on a local model.
- ClaimPilot
Insurance claims: 99.3 % of critical fields correct.
What changes in the company
AI works with sensitive data without it leaving the company, and costs do not grow with every question.
Technology and data security
Data stays on your hardware. I can replace personal data with tokens before processing, as in ClaimPilot.
Price
€1,350–3,150 per deployment, 2–4 weeks. Hardware is paid separately. Pricing
FAQ
What hardware do I need?
Smaller models run on an ordinary computer. Larger ones need a powerful GPU or a Mac with Apple Silicon. I recommend a specific setup based on your tasks.
Is a local model as good as ChatGPT?
For narrow tasks like document search or data extraction, often yes. For general reasoning the largest cloud models are still stronger, which is why I always measure on your tasks.
Is a local LLM GDPR-compliant?
A local model sends no data to a third party, which makes GDPR much simpler. Access rights and logging still need to be handled, and I set them up as part of the deployment.
How much does an AI project cost?
Private AI over your documents starts from €490, an AI audit costs €540–1,080, automating one process €800–2,400, and production RAG or an AI agent prototype €3,600–5,400. My hourly rate is €35–45 and the first 30-minute call is free.
Related
- Services
- RAG for companies
AI that answers from your documents and cites the source.
- AI audit
Where AI pays off for you and what it will cost.
- ML engineering
Custom models, fine-tuning and evaluation.