A private AI assistant that never leaves your building

Modern AI assistants running on a small computer on your own desk, with a familiar chat window, so documents and questions stay in-house.

Year
2025
Proof
Engineering guides and open-source code
Runs on
NVIDIA Jetson Orin Nano
Monitor showing a chat web interface and a terminal with jtop reporting GPU, CPU and memory use on a Jetson Orin Nano Developer Kit
Fig. 1 A local AI model answering in a chat window while the Jetson's load is monitored.

The challenge

Teams want the help of AI assistants but can't paste sensitive documents into a public cloud service.

What we built

  1. Open-weight AI models (DeepSeek-R1, LLaMA 3.2) running on a small NVIDIA Jetson computer.
  2. A familiar chat window, including asking questions about images.
  3. Memory, speed and power measured on the device before choosing a model size.

Results

  • Working setups documented step by step; the chat interface is open source.

Where this applies

  • Offices handling confidential documents
  • Sites with no or limited internet
  • Pilots before committing to larger AI infrastructure

Under the hood

  • Models served with Ollama; web UI with image input; browser-based inference with WebGPU explored as an alternative.

OllamaDeepSeek-R1LLaMA 3.2 VisionWebGPUNVIDIA Jetson Orin Nano

NVIDIA Jetson developer kit on a desk with its status light on
Fig. 2 The whole "server" - an NVIDIA Jetson developer kit.

Have a perception problem that has to run on real hardware?

Tell us what you need to detect, where it has to run and what it runs on. You'll get a reply from the founder.

Email gmujtaba@koraiio.com

Email
gmujtaba@koraiio.com
Founder
Dr. Ghulam Mujtaba
Product
DroSonic
Based in
Denver, Colorado