Documentation

Guides and references for PraetorianMind v4.0.0

Installation Guide
System requirements, initial setup, and first launch on NVIDIA Jetson AGX Orin. Covers JetPack prerequisites, Ollama configuration, and container setup.
Model Hub Usage
Browse, pull, and manage local models through the Model Hub. Includes Forge integration for custom model creation and ModelSafe archival via HammerIO.
Inference Bench Guide
Run inference benchmarks measuring tok/s, TTFT, and CUDA memory bandwidth. Track performance history and export results to CSV.
A/B Compare Guide
Set up head-to-head model comparisons. Evaluate response quality, latency, and throughput across any two locally-hosted models.
Board of Agents Configuration
Configure and govern multi-agent systems. Define agent roles, voting thresholds, escalation policies, and audit logging through the Board of Agents panel.
AegisOS NVR Integration
Connect PraetorianMind to AegisOS for surveillance feeds, recording management, motion alerts, and live camera streaming in the Security & NVR module.
AriaOS Governance Integration
Integrate AriaOS governance policies with PraetorianMind. Manage AriaFang multi-agent orchestration, compliance rules, and safety guardrails.

Quick Start

PraetorianMind runs entirely on local hardware with no cloud dependencies. Minimum requirements:

  • NVIDIA Jetson AGX Orin (32GB or 64GB recommended)
  • JetPack 6.0+ with CUDA support
  • Ollama installed and configured
  • At least one model pulled via ollama pull

Once running, access the operator interface at http://localhost:3000 and navigate to the Model Hub to verify your models are loaded. From there, explore the AI Chat, run your first benchmark, or configure the Board of Agents.

For questions or support, contact Joseph@ResilientMindAI.com.