Concepts

This section explains the ideas behind riverbank. It's for readers who want to understand the system, not just operate it.

Concept	What it explains
The compiler analogy	Why treating documents as source code makes sense
Pipeline stages	What happens at each step of compilation, including the distillation stage
Preprocessing	LLM document preprocessing (entity catalog, document summary)
Compiler profiles	The primary configuration surface
Fragment and artifact model	How documents become queryable units
Incremental compilation	Why only changed content gets recompiled
Epistemic model	All 9 status values, negative knowledge, argument graphs
Provenance model	PROV-O chains, citation grounding, GDPR erasure
Quality gates	SHACL validation, ingest gates, score functions
Multi-tenancy	RLS, named graph isolation, tenant lifecycle
Federation	Pulling triples from remote pg_ripple instances
Rendering	Generating pages from compiled knowledge