This paper introduces the Integrated Substrate Model (ISM), a layered cognitive architecture designed to address the limitations of monolithic large language models in the pursuit of Artificial General Intelligence (AGI). Built upon Anastomotic Substrate Architecture (ASA), the ISM framework proposes a modular system of interconnected cognitive layers—including reasoning, memory, emotional modulation, and meta-cognitive oversight—beneath a language interface. Rather than treating LLMs as minds, ISM treats them as expressive front-ends to deeper, structurally grounded intelligence.

This is the first public articulation of this system, developed through Hitherto Flux Hall, a division of Hitherto AI. The paper outlines the architecture’s conceptual structure, provides a symbolic diagram, and gestures toward active implementation efforts including simulated instinct, sensory-memory mapping, and recursive discernment modeling.