The hard part isn’t the screen. It’s deciding what we’re building.
How I work
MethodA brief almost never contains a problem. It contains a subject, a deadline, and somebody’s hope. The work is turning that into something a company can decide about — and then designing against the decision.
-
01
Go find the evidence
Nobody hands it to you. I go to analytics and session data, trace real journeys on real accounts, and keep asking until the question is specific enough to answer. AI lets me work through far more of it than I could by hand.
Came back with 2.2 devices per multi-line upgrade and 94% same manufacturer — the two numbers the whole solution rests on. · Multi-line upgrade
-
02
Size it honestly
Use case by use case: what breaks, which systems it touches, and which parts are not mine to solve. Saying what is out of scope is what makes an estimate believable.
Five areas assessed, three affected. Marketing and order management ruled out and said so. · Multi-line upgrade
-
03
Design against one model
One idea, held consistently, so that surfaces owned by different teams end up agreeing without anyone being ordered to.
One atomic design system shared by more than thirty newsrooms, then used to build breaking-news escalation entirely from parts they already had. · Newsroom platform
-
04
Test what you are least sure of
The load-bearing assumption first, not the safe one. With AI-built prototypes, testing it early is cheap. A study is only worth running if a bad number changes something.
The pre-configured line tested 83–88% useful — but only 62% found the way out, so we changed it before it shipped. · Multi-line upgrade
-
05
Say what actually happened
What shipped, what did not, and what the numbers did. Including the problems design could not fix.
Shipped in 45 days, then found most customers are not buying both services at once — and re-aimed the roadmap at the ones who are. · OneConnect
-
—
AI as a working method
Concepts built as working prototypes in Lovable, so research tests real alternatives. Test plans, questions and synthesis with enterprise AI. AI widens the options; research and judgment still choose.
Every OneConnect test, design, requirement and roadmap lives in one AI notebook, so each new phase starts with full context. · OneConnect
Selected work
2018–2026
About
Dallas, TXThirteen years designing e-commerce and content products, most of it in telecom, where the pricing is complicated, the systems are old, and the customer just wants to know what a phone costs.
I am the person teams call at the front of a launch, before there is anything to critique. At AT&T the saying goes that if a project is too big and too complicated, it goes to me — first in account management, now in commerce.
On every team I work with, I decide what the experience has to be. Designs change for technology, timelines and the business, and they should; my job is knowing where the experience can bend and where it cannot, because the customer would pay for it. I mentor designers, argue for research when the schedule does not want it, and care more about the pattern that outlives the project than the screen that ships this quarter.
I coach the Innovation Project for a FIRST LEGO League team — nine students, this season researching how urban heat affects native trees and local biodiversity in Dallas. It is the same work in miniature: narrow a broad theme into a problem you can actually test, go to primary sources instead of secondary reading, and get every member ready to defend it live to judges.
Also: an early-stage venture where I lead product and brand, a learning project I am building for my kids, and long hikes with no phone signal.