Knowledge Graph

Memory that
compounds.

Every conversation builds a persistent graph of entities, relationships, and patterns. lynox doesn't just remember — it understands.

Entity-aware memory

People, organizations, projects, products, locations, concepts — lynox recognizes entities and connects them automatically. "Sarah from Acme" becomes a node, linked to her company, her deals, and every conversation you've had.

Contradiction detection

"Last month you said Acme's budget is 50K. Their latest email says 80K." lynox flags conflicts instead of silently overwriting. You decide what's current.

Relationship mapping

Entities don't exist in isolation. lynox maps how they connect — who works at which company, which project depends on which deal, what pattern leads to which outcome.

Semantic search

ONNX embeddings power search across your entire knowledge base. Ask "what did we discuss about pricing" and get results from any thread, any entity, any date.

lynox
Knowledge Graph — entities, relationships, and type filters

Your competitors use RAG.
lynox builds a graph.

Retrieval-Augmented Generation searches documents. A Knowledge Graph understands structure. It knows that Sarah is the CTO of Acme, that Acme's deal is in negotiation, and that Sarah prefers calls over emails — without you asking.

Day 1

10 entities. Your first contacts and projects.

Month 1

200+ entities. Patterns emerge. lynox suggests connections you missed.

Month 6

1,000+ entities. Your business graph. Searchable, connected, yours.

Built for permanence.

Storage SQLite — no external database, runs on your server
Search ONNX embeddings — fast semantic search, runs locally
Resolution Entity deduplication — "Sarah Chen" and "S. Chen" resolve to one node
Traversal Recursive CTE queries — multi-hop relationship navigation
Languages Multi-lingual — entities in any language, cross-language search
Export Your data, your files — copy the SQLite database anytime