Primitive record types

Neotoma has seven primitive record types. They are the system-level building blocks every application-level entity, observation, and snapshot is composed from. Together they give Neotoma its determinism, immutability, and full provenance.

Primitive record types are distinct from application-level record types like contacts, tasks, transactions, and events. Application types are what agents and users write through the API. Primitive types are what Neotoma uses internally to store, audit, and compose that data.

• EntitiesCanonical record for every person, company, location, or thing Neotoma knows about
• Entity snapshotsDeterministic reducer output that gives every entity its current, provenance-rich truth
• SourcesContent-addressed raw storage for every byte that ever entered Neotoma
• InterpretationsVersioned, audited extraction attempts that turn a source into structured observations
• ObservationsGranular, immutable facts that the reducer composes into entity snapshots
• RelationshipsFirst-class typed graph edges that follow the same observation-snapshot pattern as entities
• Timeline eventsSource-anchored temporal records derived deterministically from extracted dates

How they compose

The three-layer truth model is Source → Interpretation → Observation → Snapshot. Sources hold raw bytes. Interpretations record how those bytes were read. Observations capture granular facts that link back to both, and the reducer composes them into snapshots. Relationships follow the same observation-snapshot pattern but for typed graph edges. Timeline events fall out as a deterministic side-effect of writing snapshots with date fields.

For the full architectural picture see the architecture overview and the determinism doctrine.