Rethinking AI Hardware: A Three-Layer Cognitive Architecture for Autonomous Agents
For developers of autonomous AI systems, this work introduces a hardware-aware cognitive decomposition that significantly improves efficiency and offline capability, though the evaluation is limited to synthetic tasks.
The paper proposes a three-layer cognitive architecture (Tri-Spirit) for autonomous agents that decomposes intelligence into planning, reasoning, and execution layers mapped to distinct hardware substrates. In simulations of 2000 synthetic tasks, it reduces latency by 75.6%, energy by 71.1%, and LLM invocations by 30%, while achieving 77.6% offline task completion.
The next generation of autonomous AI systems will be constrained not only by model capability, but by how intelligence is structured across heterogeneous hardware. Current paradigms -- cloud-centric AI, on-device inference, and edge-cloud pipelines -- treat planning, reasoning, and execution as a monolithic process, leading to unnecessary latency, energy consumption, and fragmented behavioral continuity. We introduce the Tri-Spirit Architecture, a three-layer cognitive framework that decomposes intelligence into planning (Super Layer), reasoning (Agent Layer), and execution (Reflex Layer), each mapped to distinct compute substrates and coordinated via an asynchronous message bus. We formalize the system with a parameterized routing policy, a habit-compilation mechanism that promotes repeated reasoning paths into zero-inference execution policies, a convergent memory model, and explicit safety constraints. We evaluate the architecture in a reproducible simulation of 2000 synthetic tasks against cloud-centric and edge-only baselines. Tri-Spirit reduces mean task latency by 75.6 percent and energy consumption by 71.1 percent, while decreasing LLM invocations by 30 percent and enabling 77.6 percent offline task completion. These results suggest that cognitive decomposition, rather than model scaling alone, is a primary driver of system-level efficiency in AI hardware.