Living Coherence is an independent research publication exploring grounded intelligence — how human and machine knowing can meet in structure.
We investigate the gap between what AI systems are given and what they need to reason correctly. The gap has a name: the absence of schema awareness in the context pipeline. This publication is about that gap — what causes it, what closes it, and what becomes possible once it is closed.
Editorial Stance
Living Coherence is an independent research publication. We are affiliated with ARAMAI — the product group of Ariesnet, Inc. that does the research and makes the software described here — but we publish as a separate editorial voice. Claims in this publication are grounded in production evidence, formal research, or clearly marked as conjecture. We do not publish vendor positioning or marketing content.
The publication draws on original research in schema intelligence, semantic architecture, and AI knowledge infrastructure. When we cite a result — like the 92% versus 68% accuracy comparison in Structured Context Retrieval: Empirical Results from Production — we mean production deployments, not controlled benchmarks optimized for press releases.
Three Lines of Inquiry
Known Shapes is ontological research: the forms that structure knowledge. What are the shapes of domains — the entities, relationships, constraints, and hierarchies that make meaning possible? This line of inquiry investigates how knowledge architecture affects AI capability, and what it means to build AI systems that understand structure rather than approximate it.
Look Up First is methodology: grounded approaches to working with AI. Before generating, retrieve. Before asserting, verify. Before reasoning, ground. This line of inquiry develops and tests the practical methods for building AI systems that are correct rather than merely fluent — with a particular focus on Structured Context Retrieval and the production evidence for its effectiveness.
The Stranger Problem is frontier research: what happens at the edge of the known. When AI systems must cooperate with other AI systems they have never encountered — across organizational boundaries, across ontological gaps, without prior coordination — how does semantic alignment happen? This line of inquiry investigates the hardest version of the distributed AI problem.
The Research Program
The methodologies described in this publication — Schema-Aware Foundation for Enterprise AI (SAFE), Structured Context Retrieval (SCR), Runtime Ontology Schema Editing Through Type Alignment (ROSETTA), OCTOPUS distributed cognition — are developed and maintained by ARAMAI. They are documented formally in the Methodology Library.
Where that research becomes software, it becomes CoreModels, ARAMAI's platform for holding the models an organization already has — data warehouses and dbt projects, CMS content types, vocabularies, standards — as one connected graph, and maintaining the mappings between them so a term can be looked up at runtime by the systems and agents that need it. The Schematica schema library is the open reference collection alongside it. We write about the problems; those are where the work of solving them happens.
The research program began with a single observation: the way most AI systems receive context is architecturally broken for enterprise use cases. Fixing it requires not just better algorithms but a different architectural stance — one in which structure is preserved, grounded, and made machine-readable rather than destroyed in the retrieval pipeline.
Contributor
Cruce Saunders is the founder of ARAMAI, the architect of CoreModels, and the primary author of this publication. Before ARAMAI he founded and led [A], a content intelligence consultancy that designed structured content and semantic systems for some of the largest enterprise publishers in the world, including Mayo Clinic, PayPal, Cisco, Adobe, and Microsoft. He is the author of Content Engineering for a Multi-Channel World, a founder of the Kinetic Information Association, and a longtime speaker on content intelligence, semantic standards, and the coherence of content, data, AI, and people.