LATTICE, not RAG
Retrieval-Augmented Generation is a hack around not-having-a-graph.
RAG papers over the wrong problem
RAG asks: our language model does not know this, so let us stuff a chunk into the prompt and hope. It works — partially, at high latency, with brittle top-k.
LATTICE walks a real graph
LATTICE — Layered Anchored Typed Traversal — walks the NEXUS graph deterministically from the query anchors. There is no similarity threshold to tune. There is a path, or there is not.
A worked example
Query: 'does my child need MR vaccine at nine months in Nepal?' LATTICE anchors on {age=9mo, vaccine=MR, country=Nepal}, walks to the EPI schedule entity, returns the yes/no plus the exact schedule row. No vector cosine. No token stuffing. A walk.
Why we use different words
Where the industry says LLM, we say CLLM. Where it says vector database, we say NEXUS. Where it says Chain-of-Thought, we say PRISM proof-tree. Where it says RAG, we say LATTICE. Where it says trillion-token dataset, we say Smart Router Dataset. Different words because different architecture. Read the whitepaper.