Skip to main content

​
Core: FalkorDB GraphRAG

Membria uses a hybrid GraphRAG approach where graph and vector retrieval are executed in a single system. Embeddings live inside graph nodes to avoid sync issues between separate stores.

​
Event schema (Wikontic-inspired)

Every decision is stored as a structured event: 
DECISION EVENT:
- WHO: Partner | Associate | Committee
- WHAT: Decision | Precedent | Matter | Document
- WHEN: Timestamp (immutable)
- WHY: Reasoning text (embedded as vector)
- CONTEXT: Client, matter, practice area
- RELATIONS:
  - approved_by -> Person
  - references -> Precedent
  - deviates_from -> Standard
  - resulted_in -> Outcome
  - linked_to -> Document
- QUALIFIERS: confidence, status, review_required

​
Integration architecture

Membria connects to the systems you already use:
  • Document watcher (extracts decisions from docs)
  • Email analyzer (detects decision discussions)
  • Meeting summarizer (captures verbal decisions)
These connectors feed the Membria core (GraphRAG, Decision Surface, DBB).

​
Deployment options

OptionDescriptionTarget
Cloud SaaSMulti-tenant, Membria-hostedSMB, quick start
Private cloudSingle-tenant, customer cloudMid-market, compliance
On-premiseCustomer infrastructureEnterprise, banks