Local AI
Also called: private AI, on-premises AI, on-prem AI
Local AI runs an artificial intelligence model on your own hardware, at the office or at home, without sending your data to the cloud.
With an online assistant, every question and every document goes to a provider's servers. Local AI, also called private AI or on-premises AI, reverses this logic: the model is installed on a computer or server that you own, and all the computing happens on site. Thanks to open models such as Llama, Mistral, Qwen or Gemma, you can now get very useful results without depending on an outside service.
The first benefit is confidentiality: contracts, customer files, medical data or technical drawings stay on your premises, which also makes GDPR compliance simpler. On top of that, there is no per-user subscription or per-token billing, it works without an internet connection, and you are free to choose and customise your models. On the other hand, the largest models on the market remain out of reach for a desktop machine, and maintenance is your responsibility unless you hand it over to a service provider.
When equipping yourself, everything starts with GPU memory. A 16 GB card comfortably runs quantized models of 7 to 14 billion parameters. With 32 GB spread across two cards, you can use more capable models or longer contexts. On the software side, Ollama serves the model and Open WebUI provides a chat interface close to that of mainstream assistants.
Our AI workstations come with Linux, NVIDIA drivers, Ollama and Open WebUI already installed. We also offer, on quotation, a private AI assistant connected to your documents.