Case Study
VML × ConEdison
Embedding AI into design and engineering workflows at scale, leading two Agile pods for New York's largest utility while building a WCAG-compliant design system and raising AI literacy across cross-functional teams.
Overview
This engagement for ConEd customers is more than just improving UX. Leading two Agile pods across both ConEdison and Orange and Rockland Electric, integrating AI-driven workflows for design and development, brand asset management as a fluid system, and providing quality experiential interactions for emerging benefits and programs.
I've been driving this parallel mission: demonstrating that design and development doesn't just consume AI tools, it leads their integration across multiple software platforms.
ConEdison customer portal — Customer Billing pod. Multi-account management, payment flows, and account snapshot modules.
Leadership Context
Managing at Every Level
Two pods. One design direction. Zero ambiguity.
Managing two Agile pods simultaneously means holding design coherence across parallel workstreams that often pull in different directions. Customer Billing and Account Management share a platform but diverge in complexity, data density, and stakeholder pressure.
- Lead design strategy and UX discovery sessions across both pods
- Manage stakeholder expectations through high-fidelity prototypes and executive presentations
- Partner with Legal and Content to ensure alignment before engineering picks up a ticket
- Mentor designers on critique structure, career growth, and quality standards
Two Agile pods operating in parallel under a unified design direction
"Stakeholder alignment at an enterprise like ConEdison isn't a meeting; it's a practice. Prototypes do the talking that slide decks can't."
AI Integration
AI in design isn't about replacing judgment. It's about compressing the distance between an idea and a testable prototype.
Workflow Integration
Embedding AI at every stage of the design process
I introduced AI tooling incrementally — starting with the highest-friction areas and building team confidence before expanding. The goal wasn't adoption for its own sake; it was measurable speed and quality improvements that both design and engineering could feel.
- Prompt-driven wireframe divergence: 4–6 layout directions in the time it previously took to produce one
- AI-assisted UX copy iteration paired directly with content designers
- Generative component variants in Figma to stress-test design system edge cases before engineering encounters them
- AI-supported accessibility checks catching contrast, label, and focus order issues before handoff
Lunch & Learns + Coaching
Raising AI literacy across two organizations
I run regular Lunch & Learn sessions for VML and ConEdison teams — designers and engineers — with live tooling demos teams can apply the same day. One-on-one coaching focuses on building critical judgment: when to use AI output, when to reject it, and how to review it with the same rigour as any other design decision.
- Live sessions covering Figma AI features, AI-aided prototyping, and prompt engineering for design
- Tailored coaching for designers at different comfort levels — skeptics to early adopters
- Engineering-focused sessions on design-to-code handoff improvements
- Documented AI usage guidelines and decision frameworks shared across both organizations
Cross-functional AI sessions bridging design and engineering at VML and ConEdison
Design System & Accessibility
Design System
A living system that engineering actually uses
Building the ConEdison design system means holding two things in tension: the creative flexibility teams need to solve real product problems, and the structural consistency that makes a platform serving millions predictable and trustworthy.
- Token-based Figma system — primitives → semantic → component — mirroring the CSS token structure in production
- Component documentation with usage guidelines, state specs, and explicit do/don'ts
- Design system contribution model preventing fragmentation across pods
- Brand guideline stewardship across web portal, mobile web, and tablet breakpoints
ConEdison design system — token architecture mapped to production CSS, maintained across two pod workstreams
WCAG 2.1 AA Compliance
Accessibility as infrastructure, not a checklist
ConEdison serves all of New York — the platform must work for users who are elderly, visually impaired, or under stress during an outage or billing dispute. Accessibility is a design constraint from the first wireframe, not an audit item at the end of a sprint.
- WCAG 2.1 AA contrast ratios enforced at the token level — no component ships with a failing color pair
- Focus management and keyboard navigation patterns documented and tested across critical flows
- Screen reader annotation layer in Figma: every handoff includes ARIA labels, landmark structure, and reading order
- Accessibility review integrated into the sprint definition of done
WCAG compliance built in at the token level — color, contrast, and focus patterns enforced across every component
"When accessibility is in the token, it's in every component that inherits it. You fix it once. It propagates everywhere."
2
Agile pods led simultaneously: Customer Billing and Account Management
22%
Improvement in core task-completion rates across platform redesign initiatives
18%
Reduction in user drop-off following platform-wide navigation and IA redesign
AA
WCAG 2.1 AA compliance built into every component in the design system from day one
What I Learned
Reflection
AI adoption is a leadership problem before it's a tooling problem
The technology is the easy part. Getting a team to shift how they work — especially in a regulated enterprise with legacy process inertia — requires sustained investment in trust, demonstration, and psychological safety.
The Lunch & Learns that landed weren't the most impressive demos. They were the ones where I solved a real problem the team had that week, live, using AI. That's the proof of concept that changes behavior.
- Design leadership in an AI era is about judgment, not tool coverage
- Making that judgment legible and repeatable is what separates a workshop from a culture shift
- This engagement continues to evolve as AI tooling and team capability compound
Ongoing — this engagement continues to evolve as AI tooling and team capability compound
Get in Touch
Like what
you see?
Let's talk about your next project.