Bill Sands
Experience Design Executive
AI Design Systems | Agentic Architecture | Design Standards as Code
Most design leaders are being asked to do two things at once right now. Keep delivery moving and rebuild the model underneath. That is not a comfortable position. It is, however, exactly where transformation happens.
I have spent the last four years turning traditional design organizations into AI-driven ones, most recently inside a complex, regulated, global platform environment, while the pressure to show AI results was already in the room.
About
The person behind the model.
I build design functions that scale with the platform, not the headcount.
For the last four years that has meant leading AI transformation inside enterprise design organizations, first across a global customer support and success function, now across a regulated financial services platform operating in 40+ markets.
The current work is an enterprise AI transformation program spanning five workstreams, a four-layer agentic design architecture, and a governed design system. Built from zero over two years. Still running. It is the most advanced version of a problem I have been solving for 15 years. The question is how organizations close the gap between what they intend to deliver and what actually ships.
Before this: design operating model transformation and experience strategy work across financial services, enterprise technology, and consumer brands. Different industries, same structural problem.
Based in Atlanta.
The Work
Building a design function that scales with the platform, not the headcount.
Currently in productionThe design function inherited was built around people. Every new market, every new product, every new channel required more designers, more contractors, more agency capacity. That model does not scale.
What was built instead: a design operating model structured around platform thinking. Governance replaces oversight. Reusable platform assets replace one-off solutions. AI augments execution rather than waiting for a team large enough to absorb the volume.
20+
Markets supported by a flat team
75-80%
Reduction in external contractor dependency
2 Years
From zero to a functioning AI-augmented global design platform
Note: You cannot stop delivery to fix the model underneath it. That is what made this hard. Everything here was built while the work was still running.
The System
The design system becomes infrastructure.
In productionFor AI to produce output that ships, it has to read from something authoritative. The shift is treating the design system less like a component library and more like an interface, the thing AI agents read from, reason over, and build against. That requires the system to be machine-readable, with tokens and patterns flowing into the build pipeline directly rather than being interpreted at each step.
The architecture runs in four layers.
Foundation
the governed baseline, design system, tokens, and components. What everything above it reads from.
Surface and bridge
the mapping layer connecting design system components to their code equivalents, so engineering receives implementation-aligned output at handoff rather than a specification to interpret.
Intelligence
research, voice of customer, analytics, and operational data made accessible through AI-assisted workflows.
Context
the layer that makes platform knowledge queryable by any agent or tool, so AI produces on-standard output rather than generic output.
Note: This direction is where the industry is heading. What is harder is running it across dozens of regulated markets with distinct compliance, localization, and layout requirements. The sequencing is the discipline. The tools live at the surface layer, which makes that the tempting place to start. Without the foundation stable underneath it, the output has to be corrected every time.
The Standard
Governing experience quality across markets without growing the team.
Active and scalingA design system exists in most organizations. Getting teams to actually govern against it consistently, across markets, across time zones, across engineering cultures, is the harder problem.
What was built: a governance model that treats the design system the way a product owner treats a product. Baseline adoption, design system coverage, automated QA coverage, and exception rate are the scoreboard.
40+
Active global markets governed by a single baseline
WCAG 2.1 AA
Accessibility compliance built into delivery gates
Automated
Design QA validating system compliance and front-end parity
Note: What makes a design system actually govern at scale is whether the decision rights behind it are clear. Who can override. What triggers a review. What counts as an exception. Getting that clarity across markets can take longer than building the system itself.
The Workflows
7 workflows, each with a measurable outcome.
In progress, first workflows operationalThe program defines seven delivery workflows where AI is applied directly rather than experimented with adjacent to the work. Each has defined inputs, outputs, an owner, and a success measure.
- Requirements to prototype
- Mobile to web generation
- Documentation automation
- Market delta generation
- Automated QA
- Design to code
- Platform contribution automation
80%
Reduction in design effort on mobile-to-web generation
7
Defined workflows with named owners and success measures
5
Program workstreams spanning Product, Design, Engineering, and Markets
Note: The measure that matters is not how many tools the team is using. It is whether the work itself changed. Reporting outcomes instead of activity is the harder discipline and the one that makes the program real.
Earlier Work
Where the operating model thinking came from.
The work below spans financial services, enterprise technology, and consumer brands. The problem across all of it is identical. Organizations that need to deliver better experiences faster, at scale, with less friction between strategy and execution. Each engagement produced both a strategic operating model and the design outputs that made it actionable.
Applied generative AI across the design value chain to improve operational efficiency by 23%, an early enterprise proof point for AI-augmented design delivery. Delivered 8% reduction in case load and 5% operational efficiency savings through journey and workflow redesign. Built a design operating model connecting customer support and success delivery to measurable business outcomes.
Chick-fil-A
Led a full experience redesign for restaurant operators and the tools they use to run restaurants at scale. Research and transformation program across 15 locations and 12 weeks. 28 interviews, 6 workshops. Produced a redesigned mobile app and global portal with new design guidelines and an operating model to support ongoing delivery. Delivered 22% cost reduction and 90% user adoption.
Mitsubishi Bank
Led experience design and operating model work for one of the world's largest financial institutions. Macro experience redesign across account management and key service areas. Built design guidelines and an experience operating model connecting product portfolio decisions to strategic objectives.
Equifax
Led research and design strategy for an organic growth engagement on a new B2B data product. Research program across 6 weeks, 120 survey responses, and 18 interviews. Generated 36M in reach, beat competitor CTR benchmarks by 38%, and drove brand favorability and lead generation.
Let's talk.
If the work above matches the problem you are trying to solve, the conversation is worth having. A 30-minute call is where it starts.
Email: wrsands@outlook.com
LinkedIn: linkedin.com/in/sandsbill