The Olmex Standard: How We Define Organizational Intelligence
Every new product category needs a definition. Not a marketing slogan --- a precise, testable standard that separates the real thing from the imitation. This is the Olmex Standard for organizational intelligence.
The Four Requirements
A product that claims to provide organizational intelligence must demonstrate all four of the following. Missing one is not a partial product. It is a different category.
1. Evidence-Gated Synthesis
Every claim in every synthesized output must trace to one or more source signals. The product must block claims that cannot be traced, route ambiguous claims through an ask loop, and omit claims that can be neither cited nor resolved. A product that guesses, omits without asking, or presents inference as fact does not meet this standard.
2. The Ask Loop
When the system encounters a signal it cannot explain, it must ask --- not guess, not omit. The ask must be routed to the person most likely to know, formulated specifically enough to answer in 30 seconds, and the answer must be stored as permanent context that improves future synthesis. A product that routes asks to a generic chatbot, that broadcasts to channels, or that does not store answers does not meet this standard.
3. Two Cadences, One Truth
The product must serve both the daily clock (awareness) and the weekly clock (judgment) from the same accumulated model. The daily brief must be a delta projection of the model, not a stateless recompile. The weekly review must be the full synthesis. The two outputs must never contradict. A product that serves only one cadence, or serves both from independent pipelines, does not meet this standard.
4. Persistent Model
The product must hold state across compiles: entities, relationships, provenance, confidence, permissions, history. The model must accumulate, not reset. Ask answers and user corrections must write into the model as first-class, high-trust edges. A product that is stateless per compile, that treats each week as an independent snapshot, does not meet this standard.
Why These Four
These four requirements are not arbitrary. They are the minimum set that makes organizational intelligence distinct from: - Search: find what you already know to look for - Dashboards: show what someone chose to instrument last quarter - Feeds: aggregate without filtering - Chatbots: respond without memory
A product that meets all four is not a better search engine, a prettier dashboard, a smarter feed, or a more conversational chatbot. It is a different category: a system that continuously builds a live model of how a company operates and delivers proactive, evidence-gated synthesis on the cadences the company actually runs on.
The Test
The Olmex Standard is not a certification. It is a test that any product can apply to itself: - Pick a recent week. Can you trace every claim in your output to specific signals? - Pick an ambiguous signal. Do you ask, or do you guess/omit? - Pick a daily and weekly output from the same week. Do they contradict? - Pick a claim from three months ago. Is the model that produced it still accessible, or was it stateless?
Four yeses = organizational intelligence. Any no = something else, possibly valuable, but not this category.
The Honest Position
Olmex meets this standard partially today. The evidence gate and ask loop are live. The weekly Operating Picture is live. The persistent model is the core build of the next two quarters. The daily brief ships after the model is proven.
We publish this standard not because we have achieved it fully, but because we are building toward it openly. The standard is the constraint. The product is the execution. The gap between them is the work we share with our customers.
A category without a standard is marketing. A standard without a product is vaporware. The Olmex Standard is the bridge between them.
The Olmex Standard is a living document. Amendments require evidence of what changed in the world to justify the change in position. Current version: 1.0, July 2026.