Atonix Digital Design System Product Design Enterprise AI / ML

DesignOI, a system
for Operational Intelligence

A design system purpose built for industrial diagnostics, started as a UI kit with static components, but evolved into a framework of token architecture that spans tools and disciplines purpose built for moving raw data to confident action.

Role
Lead Designer, System Author
Scope
System + Product, End to end
Domain
Operational Intelligence
DesignOI Alerts view

Black and Veatch kept the design. Atonix started again, from zero.

When Atonix separated from Black and Veatch, BV retained Asset360, its dark theme, and its entire color palette. Atonix kept the Monitoring and Diagnostics product and nothing else. A new visual direction was required from scratch.

Monitoring provided real-time visibility. Diagnostics delivered predictive insight and recommended action. But the two lived in disconnected experiences. Operators carried context manually between them at exactly the moment they could least afford the friction.

As code grew patterns diverged.

The same architectural slot — the right panel — solved differently across four Task Centers before DesignOI existed

The audit made the case.
The foundation was the answer.

A systematic review of every screen surfaced the same story: no button architecture existed, and engineers were solving the same interaction differently screen to screen.

Engineering agreed: Get these right and the system has something solid to stand on. DesignOI was created.

DesignOI color ramp, type scale, and button specification

Documented in two places. Kept in sync.

Each component was validated in both places before it was done. Figma for spec and Storybook for working code. Catching gaps between what was spec’d and what was built before they reached a shipped feature.

Figma and Storybook List Components

DesignOI had cross team ownership

Head of Engineering Front-End Developers Product Managers QA
EXECUTIVE SPONSORSHIP

Head of Engineering drove the call

Together with the head of engineering we recognized we were beginning to duplicate patterns. As solutions and the road map was extending it was the right time to make component, patterns, and templates real.

DOCUMENTATION

Self-service adoption

Built component docs engineers could use without a designer present. Usage rules, anatomy guidelines, and visual specs, Lowered the cost of doing it right.

Faster Development. Consistent Implementation. One source of truth.

Figma and Storybook List Components

Then: a static component library.
Now: a token architecture scaling systems and disciplines.

DesignOI shipped as a documented component library with Storybook implementation. That solved the immediate problem. But a static system still requires human maintenance, tokens drift, specs go stale, handoff breaks down.

The proof: the Operational Intelligence login screen below was designed entirely through conversation — no traditional Figma workflow, no manual spec documents. A prompt describing the intent, the DesignOI system spec, and Figma variables did the rest. Brand-compliant. Token-accurate. Production-ready.

Claude conversation alongside the DesignOI Operational Intelligence login screen designed conversationally

From design system to working code. An AI-assisted implementation.

DesignOI didn't stop at the design system. The token architecture and component-library.md became the foundation for a production React component library, built using AI-assisted tooling to close the gap between Figma and engineering permanently.

The Figma component library was connected directly via MCP, allowing Claude to read actual token values, component states, and variant structures rather than working from documentation. The result is a system where the design and the code are derived from the same source.

What was built

Deliverable Detail
Design token system Color scales, type styles, spacing, shadows. All extracted from Figma into CSS custom properties and W3C token JSON
React component library 14 components built to exact Figma specs with TypeScript APIs and Storybook documentation
Figma Code Connect Every component wired to its Figma counterpart. Developers see real import syntax in Dev Mode

Components built: Button, Badge, Input, Select, Checkbox, Switch, Card, Tooltip, Dialog, Tabs, Notification Toast/Banner, Chip, Table, Navigation Rail

The AI didn't generate generic components, it read the Figma file directly, matched exact states and variants, and maintained token consistency across the entire system. Design and code stayed in sync from the start.

What building a system from scratch teaches you.

DesignOI worked because it was designed around a specific cognitive model, not because it had a comprehensive component library. The framework came first. The components followed.

What worked well

The four stage framework gave every design decision a clear test: does this move operators from detection to action?
Building through the product meant every pattern was validated against real workflows before promotion to the system.
Replacing modals with progressive disclosure was the most impactful single decision. Operators could go deep without losing their place.
Accessibility as a system requirement prevented an entire class of defects and produced better interfaces for everyone.

What's next

Expand the system to support AI assisted recommendations as ML capabilities mature into the product layer.
Add component usage analytics to understand adoption and surface system gaps.
Adaptive interfaces that adjust density and depth based on operator expertise level.
Tighter token integration with engineering to reduce design to build drift over time.
Let's talk about your
next big idea.

I'm currently open to new design leadership roles. Whether you're building something ambitious or improving what you already have,
I'd love to connect.

Get in touch