Ziyx
2025–26
One source of truth for games, screens and developers, and the learning games and dashboard that came out of it.

Role
Senior Product Designer
Duration
July 2025 to April 2026
Services
Web
Category
EdTech
Overview
Teach business skills without it feeling like homework
Ziyx teaches entrepreneurship to young adults. Games and screens had to ship quickly without each one looking like a different product. I built the design system, then designed the games, the AI feedback and the progress dashboard on top of it.
The Impact
The result I can stand behind is a shared system, games that teach one idea each, and a dashboard two audiences can read. No percentages beyond what the case study claims.
One
source of truth
Tokens, components and documentation
AA
WCAG 2.1
Contrast, focus and touch targets by default
Two
audiences
Learners and administrators, one dashboard
Problem
Ship quickly, and still look like one product
A standard SaaS kit would have been quicker, and wrong. The games needed expressive components, and the screens needed to stay consistent with them.


01
Every screen its own product
Games and product screens were at risk of drifting apart.
02
Homework, not play
Business skills for young adults die if the lesson feels like a form.
03
Progress you cannot see
Learners and administrators needed engagement to be readable at a glance.
Research
Games first, then the system that could hold them
The system was shaped by the product: short visual games, a bright youthful brand, and a dashboard for two kinds of people.
Three games
Resource Radar, Model Matters and Customer Connect.
One system
Tokens for type, colour and icons, plus variants and documentation.
One dashboard
Progress for learners and administrators.
What we learned
Tokens first, components second. A tweak to colour or type should happen in one place. The AI feedback also needed a design for the moment it is slow or wrong.
Handoff went from guesswork to copy and paste.
Lead developer
Ziyx
Organizing recurring patterns and behaviors
The recurring split was expression versus consistency. Games needed more flexibility than a dashboard kit, so the system took on a bit more maintenance.

Key insights
Accessible by default is cheaper than an audit at the end.
AI feedback needs loading, fallback copy and a way to try again.
Each game should teach one idea, through short visual play.
Developers needed documentation and naming, not just a Figma file.
Understanding the user journey
From a new game that might not look like Ziyx, through play and feedback, to a progress view someone can read quickly.
01
Start a new game
Friction
A new game could look unrelated to the last one.
Opportunity
Tokens and expressive components, documented, with responsive constraints.
Impact
Games ship on a shared system and still look like Ziyx.
02
Play one idea
Friction
The lesson failed when it felt like homework.
Opportunity
One idea per game: resources, business models or customers.
Impact
Short visual play, iterated on clarity and difficulty.
03
Get feedback
Friction
AI prompts tailor the response, and sometimes miss.
Opportunity
Flows for tailored feedback, plus a graceful failure.
Impact
The learner can try again when the model is off.
04
Check progress
Friction
Completion buried in a list helps no one.
Opportunity
Clear hierarchy, simple data patterns and filters.
Impact
Learners and administrators read progress at a glance.
Ideation
Built for games, not just dashboards
I chose more flexibility, and a bit more maintenance, so the system could cover expressive game UI and ordinary product screens.

Token structure, a component with its variants, and a screen before and after the system.
01
Tokens first
Type, colour and icons as tokens, so a brand change ripples instead of being redrawn.
02
Expressive components
Variants for game UI, not only form controls, with auto layout and responsive constraints.
03
A dashboard on the same system
Layout hierarchy and filter interactions for learners and administrators.
Testing interaction flows early
I iterated the games on clarity, difficulty and engagement, and took part in accessibility audits and QA reviews. Next I would test the AI feedback with learners, to see whether it builds confidence and not only completion.

Designs
The system, the games, the dashboard
Components with variants, spacing and grid standards, version control and documentation. Then the games and the progress views that use them.
Resource Radar, Model Matters and Customer Connect, in one art style.
The progress dashboard: hierarchy, simple charts and filters for both audiences.
AI feedback states, including loading, a miss, and a way to try again.
Lessons
A system is only as good as the habits around it
If I were extending it, I would document each component as I build it and bring developers into naming early. The games were the reason the system had to be more than a SaaS kit.
Key outcomes
A shared system
Components, variants, tokens and documentation.
Games that teach one idea
Resource Radar, Model Matters and Customer Connect.
Feedback that can fail safely
Loading, fallback copy and a retry, designed with the prompts.
A readable dashboard
Progress and engagement for learners and administrators.

