Flagship · Pre-Launch / Internal Use

Sovereign Council

Five models vote before any code installs. No quorum, no merge.

What It Is

Sovereign Council is a multi-model code review platform that gates every code change behind voting consensus across Claude, GPT-4, Gemini, Grok, and Llama. Each model reviews from its own training-prior perspective — what one model misses, another catches. A consensus bridge resolves disagreements. Sovereign safety law injection (Patent 074) enforces hardware-level laws that cannot be overridden by any single model's recommendation.

This is the internal gate before any code change merges into Aurum apps. It runs as part of the Aurum Shield security organism alongside the nightly Red/Blue/Purple cycle, canary privilege escalation detection, and the ULTRA VIOLET counter-offensive layer. The Council is the last line of defense at the code level. The Shield handles runtime and network.

Stage

Pre-Launch / Internal Use Only. Currently used as the internal code gate for all Aurum Nebula development. Not yet packaged for external customers. Patent coverage filed across the full security stack.

Patents (Provisional — USPTO)

All patents are provisional filings through Aurum Nebula LLC. USPTO provisional → utility conversion deadline: March 2027.

Technology Stack

NIMMulti-model routing — each model served via optimized inference microservice
NeMo GuardrailsSafety direction enforcement across voting models
TensorRTPer-model inference optimization for low-latency vote cycles

Current: Claude API, GPT-4, Gemini, Grok, Ollama (Llama3 local), Node.js consensus bridge.