The pipeline
EDN turns raw content into governed context in a fixed sequence of steps.1
Content enters
Your application sends content to EDN: a memory write, a document to ingest,
or a correction. Each item arrives through a typed endpoint in the API.
2
Classification
EDN classifies each record with a provenance class that records its origin,
for example whether a user stated it, a user confirmed it, a document
verified it, or a model inferred or generated it.
3
Governed storage
Records are persisted with their type, provenance, domain, and an embedding.
Superseded records are marked rather than silently overwritten, so history
stays traceable.
4
Deterministic retrieval
When your application asks for context, EDN ranks stored records by cosine
similarity to the query, applies a fixed similarity floor, and gates the
results by provenance. No model chooses the records.
5
Governed context
The selected records are assembled into a compact package, the SIP, with a
fixed set of sections. Inferences are marked as inferences and AI-generated
content is confined to an uncertain section. Your application feeds this
package to your model.
Where the model fits
The model is only ever the consumer of the final package. EDN classifies, stores, retrieves, and assembles. Your model reads the assembled context and produces the answer. This separation is what lets EDN keep provenance intact end to end.Retrieval performs no generative model calls. The query is compared against
stored embeddings with a deterministic similarity query. See
Core concepts for detail.