GEICOCase Study · 01Hue design system2025 → Present
A modern & cohesive ecosystem.
Converging tokens, components, and conventions into a federated system of systems — and bringing a long-tenured team along with it.
Role
Sr. Manager Design Systems & Gen UI
Timeframe
July 2025 — Present Ongoing
Design system
Hue Flutter + Web Components
Focus
System of systems Modernization Team alignment
Meaningful contributions
70%
Core system
Standard components, shared language.
20%
Subsystem
User-specific patterns.
10%
Custom
Single-use design.
01 / The setup
A talented team with tokens and atoms — but not a system.
When I joined, GEICO had tokens, atoms, a Figma library, and a team that cared deeply about quality and users. What was missing was the system context that turns those pieces into UX that scales.
The operating model around them had grown idiosyncratic and inward-facing. Product teams used the system as a dumping ground for individual design aesthetics. A single designer owned the design side of the system — with no design experience outside GEICO. The PM and tech lead had also only worked at GEICO. No external reference points to calibrate against.
Meanwhile, inconsistent UX across Figma, Flutter, and Web Components undermined trust in both the system and the team behind it.
"The work isn't to build more components. It's to install a system, and bring the team along with it."
— The diagnosis that reframed everything
02 / The strategy
Four paths to success — three building off the first.
Path 1: Future-forward design assets and processes. Rebuild the token system and Figma assets to form the foundation for a modern, high-quality, intentional, federated system of systems. Serves as context for humans and AI agents alike.
Path 2: Three balanced pillars. Design, Engineering, and Enablement working as equal partners — not Design owning it with Engineering catching up.
Path 3: A system of systems. Introduce subsystems, contribution paths, and the shared language that lets many teams build without breaking the whole.
Path 4: Quality contributions. Turn the free-for-all into clarity and structure — a target mix of core components, subsystem patterns, and legitimate custom work.
03 / Three balanced pillars
A system is only successful if it's used.
Design without engineering is a spec. Engineering without enablement is invisible. Enablement without both is a Slack channel with no answers. All three, equally invested, or none of them work.
Pillar · 01
Design
Vision, standards, accessibility, tokens, components, patterns, and guidelines. The source of truth for what the product looks and behaves like.
Pillar · 02
Engineering
Flutter (mobile), Web Components, APIs, tooling, and AI agent context. The layer that turns the source of truth into things you can actually ship.
Pillar · 03
Enablement
Documentation, training, office hours, AI-context guidance, adoption metrics, and a discovery process that actually engages the teams we serve.
04 / System of systems
Turning a free-for-all into clarity and structure.
Once we introduced subsystems for user-specific patterns, and stood up strong design assets underneath them, we were in a position to apply real standards and intention to contributions.
The target mix (see the visual above): 70% core components, 20% subsystem components, 10% legitimate custom work. That ratio isn't a rule — it's a rubric. It gives teams a clear way to answer should this be a system component, a subsystem pattern, or a one-off? — instead of everything becoming a one-off by default.
With a greater understanding of systems thinking and the system-of-systems model, users started interacting with the design system with real confidence. Not "am I allowed to use this?" — but "here's where this fits."
05 / The Hue-a-thon
A two-day celebration to reset the design org around a shared vision.
Two days of trainings, workshops, and collaboration. We used it to launch the system-of-systems model with the introduction of subsystems, help design and engineering teams get started with the new assets, and teach teams to use the design system as context for AI.
72
Designers
12
Hours
11
Talks
3
Workshops
2
Figma presenters
1
Reset moment
"The Hue-a-thon was a first in my career. We've all received version updates or announcement Slacks about updates to the design system — but Hue-a-thon was a unique and impactful experience. As a newer designer to the GEICO team, it was great not only seeing the improvements but having the hands-on practice as well."
— A GEICO designer
"I left the Hue-a-thon with a lot of hope. The blood and sweat poured into all the work is hard to ignore. Everything from how the system is built, usage of tokens, the documentation… the growth and shift in Hue is exponential and really shows."
— Another GEICO designer
06 / Key investments
Where the effort really went.
Relationships
Educating without alienating.
Suggesting new ways of working to senior team members needed to be approached carefully. It was essential to focus on relationship-building first — drawing folks into the diagnosis, the analysis, and the priority-setting rather than presenting conclusions.
Quality
Foundations first.
Modernizing design assets before addressing more visible concerns meant we had little to point at for several months. Everything downstream depends on it. "All roads lead from Modernization."
Team
Upskilling through hiring.
Hiring three highly experienced system designers raised the bar fast — and shifted team dynamics into a period of forming and storming. The end result was worth the work: the team is now highly collaborative and driving change.
Evangelism
Subsystems were a new concept.
A "federated system of systems" wasn't an obvious solution to most. I've presented and explained the concept many times, in many ways. In parallel, we're teaching folks what it means to plan a scalable subsystem and bring it to life.
07 / Where we are now
Progress, and what we've learned.
Progress
Shipping and adopting.
Every designer trained on the new assets, system-of-systems model, and setting AI context.
Our engineering partners are following our lead, with us providing design guidance.
Developing assets to support designers in proposing a subsystem.
Weekly workshops and office hours providing ongoing support.
Learnings
What the work has taught us.
Evolving into a system of systems is a big paradigm shift for the GEICO culture.
If we want to scale quality instead of chaos, this shift is essential while we roll out AI.
Establishing a subsystem is a large cross-craft project that most teams aren't ready for yet.
A visual refresh on the horizon is a great enticement to bring folks on board.
08 / What I'm learning
Notes from the middle of it.
You can't install a system without also installing the trust for it.
Most of my first months were spent building relationships across a long-tenured team. The token architecture matters, but the tokens don't get adopted if the people don't trust the direction. Relationships are the substrate.
"All roads lead from Modernization."
Foundations first meant nothing visible for months. That was a real cost — but everything downstream is only possible because we spent the time to get the foundation right.
Subsystems are a paradigm shift, not a feature.
You can't ship "federated system of systems" as a slide and expect adoption. It's months of presenting the same idea in different ways, until each team can see themselves in the model.
Bringing outside reference points into a long-tenured team is delicate work.
Not "here's how the industry does it, catch up" — more like "here's a way I've seen others solve this, does any of it map to what you're already thinking?" Invitation, not correction.