How a Clinical AI Machine builds organizational context.

Healthcare organizations already possess substantial clinical and operational knowledge. The difficulty is that this knowledge is distributed across systems, schemas, reports, workflows, policies, and people. It is rarely represented as one coherent, governable model of how the organization works.

HOMER-1 is Dart's proprietary architecture for building a Clinical AI Machine. At a high level, it accepts a clinical or operational objective in natural language, studies the connected organizational context, forms and investigates hypotheses, tests candidate knowledge through realistic simulations, and promotes supported results into a living customer Topology.

The important idea is not that one model produces one answer. The system behaves more like a coordinated research process: many specialist paths operate in parallel, findings can loop through several rounds of work, and human judgment is reserved for material ambiguity rather than every routine step.

A warehouse can contain the right tables while leaving basic questions unresolved. Two teams may use the same term differently. A metric can be mathematically valid but operationally misleading. A workflow may exist in policy, in software, and in practice as three different things.

The system must learn not only what information exists, but how the organization interprets, connects, governs, and uses it.

This is the role of the Clinical AI Machine: to turn fragmented evidence into a governed context layer that can support analysis and operational action without treating any single source as the whole truth.

The public model can be understood through three layers: inputs, the Factory process, and the resulting Topology.

Topology is the customer's living map of how the organization works. It can represent definitions, entities, relationships, workflows, rules, lineage, ownership, and other validated context needed to interpret questions and operate consistently.

The distinction between the Factory and Topology matters. The Factory performs the research and validation work. Topology is the governed context that survives that work. As new objectives are completed, accepted context can make later work faster and better informed because the system does not begin from zero each time.

Strategist is not Topology, and Topology is not the Factory. They are distinct layers: the Factory builds and maintains context; Topology holds the governed source of truth; products use that source of truth for analysis and action.