What this guide covers
Team AI Box turns one NVIDIA DGX Spark into a shared browser-based local AI workspace. Team members can use AI tools, files, messaging, task status, and virtual devices without installing a complete environment on every computer.
The setup is simple: prepare a fresh Spark, verify the software environment, run the installation on the Spark itself, then open the workspace in a browser.
What is DGX Spark local AI?
NVIDIA DGX Spark is a compact system for local AI development, inference, and deployment. Team AI Box adds a shared workspace layer for local AI, image generation, OCR, transcription, translation, file sharing, messaging, and virtual devices.
Requirements
Prepare a fresh NVIDIA DGX Spark running Ubuntu 24.04 ARM64 with Docker Engine, Docker Compose v2, NVIDIA container support, KVM, Git, Python 3.10 or newer, and several hundred gigabytes of temporary storage.
Use a stable network for the first setup and installation. Do not shut down or reboot the device during system updates. Install Team AI Box on a fresh system rather than on a machine containing important business data or production services.
Step 1: Verify the environment
Open a terminal on the Spark or connect through SSH:
uname -m
docker --version
docker compose version
git --version
python3 --version
Confirm ARM64, Docker Compose v2, Git, and Python 3.10+.
Step 2: Download and install Team AI Box
Run these commands on the DGX Spark:
git clone https://github.com/weerocksllc/teamaibox.git
cd teamaibox
sudo bash install/from-source.sh
The first build may download models and dependencies, build containers, prepare an offline package, and install the workspace. Do not run the commands on a normal laptop. See the Team AI Box deployment guide.
Step 3: Open and test the workspace
After installation, open the address provided by the deployment process in a browser. Test the AI assistant, document editing, file processing, OCR, translation, image generation, and task status.
External model services are optional. Privacy depends on the models, services, files, integrations, and network settings you enable.
Step 4: Configure team access
Create accounts and permissions. Team AI Box can share AI assistants, messaging, files, OCR, transcription, translation, and virtual Android, Windows, or Linux devices.
Read:
- How to Let a Team Share One DGX Spark
- What Can NVIDIA DGX Spark Do?
- Using Windows 11 ARM and Ubuntu on DGX Spark
Frequently asked questions
Can DGX Spark local AI run without the internet?
Initial setup and installation normally require internet access. After models and dependencies are installed, some local workflows can run without a continuous connection.
Can a team share one DGX Spark?
Yes. Team AI Box is designed for shared browser access. Performance depends on the model, workload, memory, and number of simultaneous users.
Is Team AI Box an NVIDIA product?
No. Team AI Box is an independent open-source project from Weerocks LLC. It is not an official NVIDIA product or certification. The project’s source and license information are available on the Team AI Box website.
Conclusion
Team AI Box lets a team set up DGX Spark local AI on one device and access it through a browser. Visit the Team AI Box deployment guide or contact teamaibox@weerocks.llc.