Whatever you run — agents, models, LLMs, custom GPTs — it was trained on a world that was only partially documented. The Unknown Unknown finds what your AI is missing and turns it into engagement, quality and decisions your competitors can't match.
Missing data is missed money. Women's economy is just the clearest example — half the global market, left on the table every single day.
The outcomes above share a single root cause: what your AI was trained on — and what it never saw.
Bias implies something present but distorted. What we are dealing with is more fundamental: centuries of knowledge, experience, and intelligence that were never entered into the record in the first place. Women who published under male names. Research that was not funded. History that was told but never archived.
The belief that "everything is online" is one of the most dangerous assumptions in technology today. It is not. What is online is a highly selective record — and building AGI on top of that record and calling it general intelligence is, frankly, naive.
The cost is already landing: Gartner expects 60% of AI projects to be abandoned through 2026 for want of AI-ready data.
The woman-man divide is also just the beginning. Gender is one axis of absence. Minority voices, non-Western knowledge systems, languages without digital infrastructure, communities whose histories were never institutionalised — all missing. All consequential.
The Unknown Unknown begins with women — the first known example of how large the missing-data problem is — and is built to expand to every gap that matters. The goal is not separation; it is completion.
"Everyone is fixing the model.
Nobody is asking what's not there."
Observability, data quality, synthetic data, governance — we build the layer underneath.
Revenue is not a side effect of the mission — it is what funds it, sustains it, and lets us give back structurally to the contributors and institutions who make it possible. The Unknown Unknown is being built as the measurement and data layer for what AI is missing. Women's economic reality is our first proof vertical — the clearest, biggest known example of the problem's scale — but the focus is every missing gap that makes your business cases less profitable. The use cases are commercially real and immediately addressable.
We are not building another AI application.
We are building the infrastructure layer that every AI system will eventually need.
The current version was built on incomplete material. Everyone who depends on it — every business, every healthcare system, every education platform — is working with a partial map.
Are you with us? We will need all the help we can get.
Join the founding