This publication defines the architectural framework of the LivingAI system and establishes formal prior art for a deterministic governance architecture for autonomous intelligent systems. The architecture introduces a governance-first execution paradigm in which probabilistic reasoning components cannot directly authorize system actions. Instead, all actions must pass through a deterministic governance layer responsible for invariant validation, authority resolution, and execution authorization. The LivingAI architecture structurally separates cognition, governance, authority, execution, and auditability into independent architectural domains. This separation ensures that probabilistic reasoning processes remain isolated from the deterministic authority and execution mechanisms that control real-world system actions. The document also introduces mechanisms for deterministic decision reconstruction (WHY-TRACE), governance routing, invariant enforcement, and containment of behavioral drift in intelligent systems operating in autonomous or semi-autonomous environments. Unlike conventional model-centric artificial intelligence systems, LivingAI implements a governance-first system architecture where probabilistic cognition is structurally constrained by deterministic governance and authority layers before any system action can occur. This publication intentionally omits proprietary implementation details while defining the architectural identity, structural model, and governance principles of the LivingAI architecture. This publication establishes formal prior art for the LivingAI architectural framework and its governance-first paradigm for autonomous intelligent systems.
Ondrej Škultéty (Wed,) studied this question.