Nexa IntelligenceMETHODOLOGY
How a fifteen-to-forty-five-minute conversation becomes a typed, dated, segment-scoped fact.
The platform runs three stages. Gather originates the raw material through AI-guided structured interviews with the people who hold the knowledge. Codify converts every accepted conversation into ten typed dimensions in a bi-temporal store. Monetize delivers the resulting intelligence to an internal dashboard, a scoped brand-partner API, and an AI-agent feed that grounds every answer on the same corpus.
This page documents the science underneath those three stages. Most of the depth lives in the first two. The conversation agent is the only instrument we have found that runs a research-grade interview at three cents per enriched conversation and holds the same protocol on interview number one and interview number ten thousand. The intelligence matrix is what lets one corpus serve an internal dashboard, the brand partners on the shelf, and an AI-agent feed without the three answers drifting away from each other.
The pages below cover the two centerpieces in depth. The Codify operational view lives on Platform. The Monetize delivery surfaces (dashboard, brand-partner API, agent feed) live on KAAS. The mechanisms behind confidentiality, anonymization, and the no-training contracts live on Trust & Security. Every claim below ties back to a mechanism you can name, audit, and replay.
THE INTELLIGENCE MATRIX
The Ten-Dimension Enrichment Schema
Every accepted conversation decomposes into ten typed dimensions. The set is vertical-stable in shape: every vertical gets ten dimensions, with two or three replacements depending on what the workforce actually knows. The page lists the ten, explains why ten and not five or twenty, and describes the pre-enrichment gate that runs before any LLM call.
Read more →Bi-Temporal Validity
Every fact carries three timestamps: learned_at, valid_at, invalid_at. As-of replay lets a brand partner run today's query and last May's query against the same corpus and get the correct answer for each. Per-dimension half-life calibration tells the system which facts decay in weeks and which decay in quarters.
Read more →The Segment Vector
Bi-temporal answers when a fact is true. The segment vector answers for whom. A surface claim about a brand can be true for Tier-4 customers today, contested at Tier-2 today, and was true at Tier-1 in 2023. The same mechanism is how one corpus can serve every brand partner on the shelf at scale without leakage between them.
Read more →Refresh Economics
Staleness is a calibrated property of the fact lifecycle. The decay function combines salience, recency, conflicting evidence, and a per-dimension half-life. The dashboard surfaces the staleness clock to the operator, and the Refresh tier converts that signal into a commercial event.
Read more →WHAT THIS PAGE INTENTIONALLY DOES NOT COVER
Codify (the productize stage) and Monetize (the distribution stage) are real, and they each have a real methodology. We document them where the operator looks for them. The Codify pipeline (pre-enrichment gate, five LLM calls, deterministic scoring, integrity controls) lives in detail on Platform alongside the operational surfaces it feeds. The Monetize distribution model (the three endpoints, the brand-partner API, the AI-agent feed) lives on KAAS alongside the pricing tiers and the proof cases.
Methodology is where you go to understand the science under capture and the data model under intelligence. That is the page you are reading. The rest of the chain is on the pages built for it.
THE ENGINEERING SOURCE OF TRUTH