Lumin Digital • 2021 - Present Google • 2020 - 2021 BMW • 2018 - 2020

Building Foundations for AI at Scale

How I unified design across Google's experimental AI and ML teams, creating foundational patterns that would scale from prototype to production as these products matured.

50M+ Future Scale
5 ML/AI Teams
85% Pattern Adoption

The Challenge

Google's emerging AI products division was brilliant at engineering but struggling with design coherence across teams.

Five different teams. Five inconsistent approaches. Limited component sharing. Users were confused when moving between products, and the brand experience was fragmented. Each team had developed their own patterns, components, and visual language in isolation.

As an engineering-led organization, design often took a backseat to functionality. We needed to prove that great design could enhance their technical excellence, not slow it down—while navigating the ambiguity of emerging AI products that were still finding their market fit.

Research & Discovery

Understanding the fragmentation before building unity.

Foundation for Future AI Products Design Tokens Colors Typography Spacing Motion Elevation Base Components Buttons Inputs Cards Modals Tables Navigation AI/ML Patterns Confidence UI Data Pipelines Model States Training UI Feedback ML Product Templates Model Dashboard Training View API Config Results View Foundation Building Blocks ML Patterns Experiences 67 Foundation Components 5 ML/AI Teams 85% Adoption Rate

Scalable architecture supporting 67 foundation components across 5 ML/AI teams

Component Audit Findings

23
Different button styles
5
Conflicting approaches
Limited
Component sharing
👥

Stakeholder Interviews

25+ interviews with ML engineers, designers, and PMs to understand pain points and workflows specific to AI products.

📊

Component Inventory

Catalogued UI elements across 5 ML products, identifying patterns unique to AI interfaces.

🔍

Workflow Analysis

Shadowed teams for 2 weeks to understand how ML engineers interact with their tools.

📈

Metrics Baseline

Established measures for design velocity, consistency, and developer satisfaction in AI context.

The Solution

A unified design foundation that celebrated engineering while elevating the user experience.

🎨

Visual Language

Created a cohesive visual system that worked across all ML products while allowing for product-specific needs and emerging AI capabilities.

🧩

Component Library

Built 67 foundation components that engineers could easily implement without sacrificing their development velocity or innovation speed.

📐

Design Principles

Established clear principles that guided decisions and helped teams make consistent choices independently in ambiguous AI scenarios.

🤝

Cross-Team Process

Created collaborative workflows that brought ML engineering and design together from concept to launch, crucial for experimental products.

Design Process

A systematic approach to unifying disparate design languages over 14 months.

(Timeline condensed - actual implementation was 14 months)

🔍

Audit

Month 1-2: Catalogued 23 button variations

📊

Analyze

Month 3-4: Identified ML-specific patterns

🎨

Design

Month 5-8: Created modular system

🧪

Test

Month 9-11: Validated with teams

🚀

Deploy

Month 12-13: Rolled out to 5 teams

📈

Measure

Month 14: 85% adoption, 45% faster

The Journey

Leading through ambiguity to create order from chaos.

Month 1-2: Discovery & Alignment

Interviewed 25+ ML engineers and designers across teams. Discovered deep mistrust between design and engineering. First design review was brutal - "This will slow us down by months."

Month 3-4: Failed Prototype

Created rigid design system. Teams rejected it immediately. Realized I was solving the wrong problem - they needed flexibility for experimental AI features, not rules. Pivoted to modular approach.

Month 5-8: Building Trust

Embedded with ML engineering teams. Learned their workflows, pain points. Co-created components WITH them, not FOR them. First successful implementation in AutoML team.

Month 9-11: Scaling Success

Other teams saw AutoML's velocity increase. Voluntary adoption began. System evolved based on real usage. Engineering became biggest advocates.

Month 12-14: Full Implementation

Rolled out across all 5 teams. Established design review process. Created self-service tools. Teams shipping 45% faster with better quality.

The Transformation

From chaos to coherence in 14 months.

Button Component Evolution

V1: Initial Audit Month 1-2 Team A Team B Team C Team D Team E 23 variations found V2: Standardization Month 5-8 Small Medium Large 3 standardized sizes V3: Final System Month 12-14 Primary Secondary Text Button Disabled + 1 unified system Discovery Iteration Launch

3 iterations over 14 months to achieve consistency without sacrificing flexibility

Before: 5 Design Languages Buttons Team A Team B Team C Team D Team E Input Fields Placeholder text Different style Underline Cards Typography Team A: Arial Team B: Helvetica Team C: Roboto Team D: System Team E: Inter Color Palettes (Variations of Google Colors) After: 1 Unified System Buttons Primary Outlined Text Disabled + Input Fields Label Input text Label Placeholder Cards Action Action Typography Scale Display Headline Body Caption Unified Color System Primary Neutral

From 5 design languages to 1 unified system

Before
✕ 5 inconsistent approaches
✕ Limited component sharing
✕ Siloed team decisions
✕ Inconsistent user experience
✕ 2-week design cycles
After
✓ Shared design foundation
✓ 67 foundation components
✓ Weekly design syncs
✓ 85% core pattern adoption
✓ 1-week design sprints

Design Leadership

Influencing without authority across a 12-person cross-functional ML team.

🎯

Strategic Alignment

Presented design system ROI to VPs. Showed how consistency would accelerate AI feature development. Secured buy-in by speaking their language: velocity and scale.

🏗️

Systems Thinking

Designed components that could adapt to unknown AI capabilities. Built flexibility into the system for features we couldn't yet imagine.

🧑‍🏫

Team Development

Mentored 5 junior designers in systems thinking. Created "Design System Champions" program. Designers became force multipliers across teams.

📊

Measuring Impact

Established metrics beyond pixels. Tracked development velocity, user task completion, support tickets. Data proved design's business value.

The Unified Visual Framework

Blue
Red
Yellow
Green

The system balanced Google's iconic colors with sophisticated neutrals, creating a professional yet approachable aesthetic that worked across all ML products while accommodating emerging AI interfaces.

The Impact

Measurable improvements in both user experience and team efficiency.

85% Core Pattern Adoption
45% Faster Prototyping
Unified Foundation Achieved
67 Foundation Components
"This design foundation let us experiment rapidly with AI interfaces while maintaining quality."
— James C., Design Lead, Google Cloud AI

Evolution & Impact

Components designed in 2020-2021 evolved into patterns now used across Google's AI products. The confidence indicators we prototyped appear in Bard, the pipeline visualizations influenced Vertex AI, and the adaptive layouts support various AI interfaces today.

Key Learnings

What building AI foundations taught me about design at scale.

Start with empathy, not enforcement.

Understanding why teams made different choices was crucial to finding solutions that worked for everyone. Authority comes from value, not title.

Make it easier, not mandatory.

The system succeeded because it made teams' lives easier, not because it was required. Adoption through attraction beats compliance.

Embrace engineering culture.

By speaking their language and understanding their constraints, design became a valued partner in ML innovation.

Design for the unknown.

Building flexibility into the system allowed it to accommodate AI features we couldn't yet imagine. Ambiguity is a design constraint.

Next Case Study → BMW: Smart Charging Innovation