Not a Chatbot. An Ontology Engine.
LLMs are great at writing poetry. They are terrible at legal accuracy. Linecite uses a graph-based ontology to ensure every fact is traceable.
Simplified view of the Linecite Extraction Pipeline
The Ontology Model
Linecite models each case as a knowledge graph consisting of Documents, Pages, Events, Providers, and Billing Records.
A "Chronology" is simply a projection of this graph: we sort Events by date and render them. This means we can re-project the same data into a "Specials Summary" or a "Provider List" without re-reading the documents.
Missing Record Detection
Because we understand the graph, we can find holes in it. If we see a "Referral to Dr. Smith" event on Jan 1st, but no records from Dr. Smith, our graph algorithms flag a "Missing Record" gap.
This is impossible for standard LLM summarizers, which only look at the text in front of them.