Lumin Digital · Principal Product Designer · 2021 – 2025

AI that speaks your design language

How I helped transform digital banking for 127 credit unions with AI-powered design systems, conversational interfaces, and intelligent automation — serving 2.1M members.

2.1M
members served
127
credit union implementations
90%
conversational onboarding completion

The challenge

Credit unions were bleeding members to digital-first banks. They needed enterprise-grade modernization without losing their personal touch.

Traditional banking UX wasn't cutting it. Credit union admins struggled with legacy tools, while members abandoned onboarding at a 73% bounce rate. Design teams spent weeks customizing each implementation, and support costs were unsustainable.

We needed something radically different — an AI-powered platform serving both B2B clients (credit unions) and their B2C members, scaling across 127 unique implementations.

The solution

Three breakthrough innovations that redefined enterprise banking UX.

🤖

Design Companions

AI assistants that understood our design system — vector stores containing the entire design language, process SOPs, and component library. Credit union teams generated brand-compliant designs in minutes, not weeks. Sprints ran 43% faster.

💬

Conversational banking

Replaced wizard flows with natural conversation. Members opened accounts like they were chatting with a friend — the AI adapted questions to responses, making compliance feel effortless. Completion rates hit 90% against a 73%-bounce legacy baseline.

🏦

Enterprise AI platform

A B2B system credit unions customized without code. The AI learned from patterns across all implementations, suggesting optimizations and automating routine work. What took 40+ hours happened in minutes — $2.3M in new revenue.

Before / after — account opening

The legacy wizard lost 73% of applicants. The conversational redesign asked the same compliance questions as a chat — and completion hit 90%.

lumin · conversational-onboarding.figDrafts · Design
68%
beforeApplication wizard — legacy
Step 3 of 9
Social Security Number
Employer name
Annual income
Employment start date
73% abandon before step 9
Back
Continue
afterConversational — shipped
Nice to meet you, Maya! What kind of account are you opening today?
CheckingSavingsBoth
Both, please
Great choice. I'll need a couple of quick details for the government — takes about 2 minutes.
90% completion · compliance intact
Conversation components
Quick reply+ 2 more
Secure field · SSN masked
Identity verified ✓
Recreated for this portfolio — original files remain on Lumin Digital hardware under NDA. Flow structure and copy match the shipped product.

The process

From ambiguous problem to shipped solution in 11 months.

Months 1–2

Enterprise discovery

Interviewed 47 credit union executives and 200+ members. Found the core tension: credit unions needed enterprise tools, members wanted personal touch. Initial wireframes failed — too corporate.

Months 3–4

Failed first attempt

Built a traditional B2B dashboard. Credit unions hated it — “feels like the software we're trying to escape.” Pivoted to studying their actual workflows.

Months 5–6

AI breakthrough

Realized AI had to absorb the customization complexity. Built the first Design Companion prototype — credit unions could finally express their brand without technical knowledge.

Months 7–8

Conversational pivot

Member testing revealed form fatigue. Experimented with chat-based onboarding. Compliance resisted until we showed 90% completion rates.

Months 9–11

Scale & launch

Rolled out to the first 10 credit unions. The AI learned from each implementation — by month 11, onboarding new clients took hours, not weeks.

Design leadership

🤝

Cross-functional alignment

Led a 12-person team across design, engineering, and AI/ML. Established a weekly “AI Design Sync” — 40% fewer revision cycles.

📚

Mentoring AI-first design

Trained 8 designers in prompt engineering and AI tools. Created the internal “AI Design Playbook” adopted company-wide.

🎯

Strategic product decisions

Convinced the C-suite to invest $1.5M in AI after presenting the prototype; defined a roadmap balancing innovation with compliance.

🔬

AI-assisted research

Analyzed patterns across 50,000+ support tickets, uncovering insights human research missed and cutting support volume 67%.

The impact

member engagement increase
$2.3M
revenue generated
47%
mobile adoption
89%
member satisfaction

“Gabe didn't just design a platform — he reimagined how credit unions operate. The AI integration feels like magic, but the real innovation is how human it feels.”

— CEO, partner credit union

Key learnings

B2B2C requires dual empathy.

Success came from making both credit union administrators and their members feel powerful.

AI scales through patterns, not rules.

Design Companions worked because they learned from usage patterns across implementations, not rigid templates.

Conversation beats configuration.

Natural language revealed user intent in ways forms never could — while making compliance feel effortless.

Enterprise AI needs human guardrails.

Automation accelerated everything, but human oversight ensured quality. The best AI amplifies human judgment.