Configure your private AI in 2 minutes. One command deploys the LLM, guardrails, knowledge base and Hermes agent, entirely on your DGX Spark, in your office. Zero cloud. Zero upload.
The problem
GDPR, sovereignty, professional secrecy: your data must not leave.
Building a central agentic AI platform takes 12 to 18 months.
Buying a DGX Spark is easy. Configuring, securing and wiring it up is not.
How it works
Describe your business need in a few clicks. No data — just a configuration recipe.
Order your NVIDIA DGX Spark via Stripe. Project-level budget, no IT sign-off.
NVIDIA DGX Spark
$4,699
Project-level budget
No IT sign-off required
On the device, the Hermes agent does it all: model, network, guardrails, knowledge base.
Your team gets a private assistant over your documents, 100% local.
Data sovereignty
Our web app only produces a configuration recipe (a .yaml file). It never sees any of your documents. The DGX Spark cooks the ingredients locally.
No external provider. The model runs on your DGX Spark, on your internal network.
Your documents never leave the device — not even during setup. The KB is built locally.
NeMo Guardrails tuned to your scope: PII masking, off-topic blocking, logging.
Configure your profile in 2 minutes. Watch your recipe build live.
Configure my AI