Skip to content
Staircase

Data

The data substrate: the graph store, the translation layer, the rules engine, the models and the documentation platform.

Where does the canonical record live, and what shapes it?

Everything above this family reads and writes through it. The graph store holds the canonical record with immutable state, so every version of an entity survives and history is a query rather than an audit table. The translation layer maps any caller's own field vocabulary onto the canonical model in both directions. The rules engine expresses the conditions that eligibility and validation test, as data rather than as code.

The remaining products serve the surfaces around those three: document and image models, the content store, the human-review queue, and the documentation platform that generated the developer reference from the ontology.

Products

In the order the value chain runs.

  1. Content has a recorded specification
  2. Language has a recorded specification
  3. ML has a recorded specification
  4. Persistence has a recorded specification
  5. Rule has a recorded specification
  6. Site has a recorded specification
  7. Turker

Type to search.