AI Financial Advisory, End-to-End Design
RyomaAI is a conversational AI platform built for Japan’s financial services industry. It guides consumers through discovery, education, and qualification, then hands them to a sales team, ready to close.
I joined as the founding and sole designer. There was no brand, no design system, no marketing presence. I built everything from zero — across product UI, brand, marketing, research, and go-to-market materials — while the product and strategy evolved around me.
Financial services in Japan faces compounding pressures: an aging workforce of insurance agents, consumers disengaged by jargon-heavy products, rising compliance costs, and digital channels that generate reach but not understanding. RyomaAI’s answer was a conversational AI that could guide consumers the way a trusted advisor would — at scale.
Designing for that required more than good UI. The product needed to feel human, adapt to different client brands, and work for a market where trust is earned slowly and financial anxiety runs high.
Before designing for Japanese consumers, I needed to understand them and the visual and cultural context they inhabit.
I built five detailed user personas representing the Japanese financial services market, varying by age, financial literacy, trust disposition, and comfort with technology. I also researched Japanese-specific design patterns, color theory, typography conventions, and popular financial services websites. Japanese digital design has distinct conventions that differ significantly from Western defaults. Designing without that context would have produced something that felt foreign to the very users we were trying to reach.
The research informed everything that followed: the product’s conversational tone, the avatar system, the marketing site messaging, and the decision to lead with empathy over features at every touchpoint.
The core product is a conversational AI interface that guides users through financial product discovery. When I joined, engineering had built a functional prototype — the visual design, interaction model, and brand avatar system did not yet exist.
What I inherited: a functional build that proved the concept but had no visual design, interaction model, or brand avatar system — the raw material I’d shape into a product.
My first month. I designed the initial UI, shaped the micro-interactions, and solved a specific challenge: the platform needed fully custom brand avatars for each client. The AI doesn’t just speak as Ryoma — it can represent a client’s brand entirely, from a samurai figure to a stylized character to a human representative. That required a flexible system that could hold radically different visual identities without losing coherence.
After a year of customer presentations and real user feedback, a clearer picture had emerged of how Japanese consumers experience financial conversations — cautiously, preferring to feel guided rather than sold to. The 2026 update reflected that learning: refined interactions, stronger visual hierarchy, and a more considered handoff between AI and human advisor.
“This wasn’t a visual refresh. It was the product catching up to what we’d learned.”
The chat app was one surface in a broader product ecosystem. I designed each touchpoint to extend the same brand system and support the same user journey.
The site needed to serve two audiences: Japanese financial services institutions evaluating a partnership, and global investors evaluating an opportunity. I wrote the full IA and copy first: structure, voice, narrative hierarchy, and per-page copy, presented across two rounds of team review before a pixel was designed. Then I prototyped the full design in Figma Make and coded the production site with Claude. Concept to launch-ready in under a week, solo, no engineering lift.
I designed two core decks and built a presentation system around them — a style guide, component library, and layout templates. Then I trained Jonathan Epstein (Chairman) and John Bennet (CEO) to use Figma so they could build and update decks independently.
When the Habito pitch came in with a 24-hour turnaround, I delivered 8 domain-expert slides as part of a 20-slide deck — half the remaining slides built by team members using my templates.
“I will never cease to be amazed at how you turn things into these amazing slides.”
Mathew Shea, CTO
A documented style guide — color, typography, shadows — plus a component library and layout templates. The system is what let non-designers produce on-brand decks without me in the room.
Founding-team design that took a bold idea from blank page to a fundable, multi-platform product — solo.
Marketing site from concept to live, no engineering lift.
Decks produced from a system I built and taught.
Of the AI Chat application, informed by customer testing and real user feedback.
Every surface: product, brand, marketing, research, presentations.