An Intelligible Vision

We’re living through the frontier moment of AI. Our forefathers explored the planet, our children may explore the stars; today, our frontier is reasoning. But frontiers do not stay open. They get explored. Then settled, because whoever arrives first sets the terms everyone else inherits. The design choices we make now will echo through the institutions of the next century.

A central question in this frontier is the role of human reasoning. Should we surrender it to superhuman AI agents?

To hand over understanding entirely to AI systems would be both a moral abdication and an operational failure. Domains like healthcare, finance, and governance already operate under bias, uncertainty, and partial observability; if we bury complexity inside black-box models, we don’t solve it; we hide it.

At Intelligible, we’re building infrastructure for automated data science: tools that connect data to foundation models while preserving the structure that guides human reasoning. We believe AI systems should be able to act on structured concepts as well as embeddings, and thus data compression is a first-order citizen of importance. Compressed data components can be human-interpretable, and are portable across model architectures and vendors.

With modular components, we enable modular systems, and modular systems encourage ecosystems. Anyone can inspect a component and adapt it. Black-box architectures wall that off, and institutions that need to outlast their vendors cannot afford it.

We’re starting in healthcare, where the need is immediate, the data are rich, and the risks of blind automation are personal. But healthcare is only the beginning. The same opportunities arise in every domain where decisions matter and systems scale.

So we are building. We’re building to advance what’s intelligible and auditable.

Let’s build.