CortexLab V6

Intelligence is a pipeline you can inspect.

Cortex V6 is not presented as a mysterious local LLM. It is an owned intelligence core built from routing retrieval graph reasoning calibration memory planning evaluation and human review. Each subsystem has a specific job and failure boundary.

146knowledge nodes
8208graph edges
192Dlocal semantic vector
0external model APIs
Ten-stage intelligence core

Every stage has a reason to exist.

01

Classify

deterministic intent rules + local neural intent model + local neural topic classifier + difficulty classifier

02

Retrieve

BM25 and TF-IDF lexical evidence plus a local dense semantic hash encoder

03

Connect

knowledge graph expansion brings prerequisites contrasts and related concepts into candidate retrieval

04

Rerank

lexical semantic graph scope exact-match and priority signals are combined into a hybrid score

05

Calibrate

confidence uses evidence agreement source coverage and top-result margin instead of copying a raw similarity score

06

Plan

Cortex selects a visible response plan based on intent topic difficulty memory availability and confidence

07

Answer

grounded extractive/template response with numbered evidence references and response-policy selection

08

Evaluate

answer quality checks grounding citation coverage confidence and uncertainty handling

09

Remember

signed-in learners can store visible deletable technical goals preferences and weak-topic memories

10

Improve

feedback updates response-policy Q-values while reviewed human teaching updates factual knowledge separately

What V6 does not claim

Architecture before marketing.

Not a pretrained frontier LLM

The dense semantic encoder is locally implemented feature hashing. It is useful for semantic-ish retrieval but it is not equivalent to a pretrained embedding model.

Not hidden chain-of-thought

Cortex exposes a compact response plan and evidence checks rather than claiming to reveal private internal reasoning.

Not automatic truth

Graph edges and human corrections influence retrieval only after scoring and review. Confidence is a calibration signal not a guarantee.