AI sovereignty local intelligence zero compromise

Offline Intelligence. Private by Design.

Run powerful local AI and secure document search—without the cloud. Your data, your hardware, your rules.

No trackers No telemetry Open models
Sentinel Console

Local chat (offline)

Apertus 8B Mistral 7B RAG (Qdrant)

Documents

Engineering Handbook.pdf Indexed
Legal - NDAs.zip Indexed
Research Notes.md Indexed

Power

CPURyzen AI 9 HX 370
NPU80 TOPS
RAM64 GB
Power~45 W

How it works

From unbox to answers in minutes. No accounts. No cloud.

01

Plug

Connect Sentinel to power and your LAN.

02

Process

Add documents. Models and index build locally.

03

Chat

Ask questions. Get grounded answers.

04

Private

Your data never leaves the device.

Choose your Sentinel

Hardware and software included. One-time payment.

Beta (Jetson)

Jetson Orin Nano 8 GB

$800
  • Models: Up to 7B parameters
  • Response speed: 2-6 tok/s
  • AI performance: ~20 TOPS
  • Support: Community forum + docs

Basic (UM790 Pro)

Minisforum UM790 Pro (Ryzen 9 7940HS + 64 GB RAM)

$1,800
  • Models: Up to 14B parameters
  • Response speed: 8-20 tok/s
  • AI performance: ~30 TOPS CPU/iGPU
  • Support: Email (48h response)
Popular

Advanced (AI X1 Pro)

Minisforum AI X1 Pro (Ryzen AI 9 HX 370 + 96 GB RAM + Radeon 780M)

$3,000
  • Models: Up to 30B parameters
  • Response speed: 15-30 tok/s
  • AI performance: ~80 TOPS NPU + GPU
  • Support: Priority (24h response)

Pro (DGX Spark)

NVIDIA DGX Spark Edge (Grace CPU + Hopper GPU, 128 GB RAM)

$10,000
  • Models: Multi-70B
  • Response speed: 80-120 tok/s
  • AI performance: ~500 TOPS
  • Support: Dedicated setup + on-call engineer

Ready to own your AI?

Configure your Sentinel and submit an inquiry. We'll confirm availability and shipping.

  • Secure submission. No trackers.
  • We reply within 1 business day.
  • You can also email us: hello@custodia-privacy.com
Back to pricing