Overture gives design intent a structure agents can understand, without flattening the work into a rigid template.

It creates a shared language for the system - tokens, components, patterns, rules, and rationale - so decisions can move between design, code, tools, teams, and the canvas of our choice.

But it stays open. Every brand has its own soul and behavior. Every team has its own way of working. So the system has to be structured enough for agents to read, and flexible enough for designers to keep shaping visually when we need to.

For us, JSON and YAML form the connective layer: machine-readable structure that tools and agents can parse, evolve, and keep in sync. That is where agents become useful. Not as a shortcut around design judgment, but as a way to help the system respond.

DESIGN_SYSTEM.MD carries the human layer: the foundational context, rationale, usage guidance, and design principles that help people understand not just what the system is, but why it works the way it does.

We’re still early, but that’s the point. The work is not just the design system. It is the way the design system gets made, used, and evolved.

This effort started as a very personal project, born from our own frustration trying to hold a more holistic, longitudinal view of our health in a system that was never designed to hold the whole story.

Health Atlas turns the records we already have into a personal health repo we can actually use. AI helps organize the details, surface patterns, notice what has changed, and prepare better questions for the next appointment.

The question we’re trying to answer is simple, and one many people are also asking: what if a clear picture of our health didn’t have to live behind expensive concierge medicine memberships or fragmented hospital portals?