Designing a Gen Z mobile brand, from research to build-ready in one quarter

Circles.Life’s platform could run VNPT’s new brand. What it couldn’t tell us was why a young customer would leave the carrier they already had. So before anyone designed a screen, I set up discovery in Hanoi with VNPT’s product managers.

I wrote a survey for young mobile users, and 171 answered. It showed us where people got stuck, but not why. So I turned those findings into an interview guide, and we ran 12 interviews. They changed our minds on onboarding: going digital wasn’t the fix, explaining each step was.

We knew what Gen Z wanted. Now we had to design it before any PRDs existed, while an outside agency was still creating the brand. I set the direction, split the journeys between two designers and myself, and stepped in wherever a journey was critical or slipping.

Waiting for PRDs would have cost us the quarter. And three teams were working across three languages: VNPT in Vietnamese, KDDI in Japanese, and our team in India in English. Long documents got lost in translation; a screen didn’t.

So I moved design upstream. We sketched low-fidelity wireframes and built clickable prototypes in Figma Make, then used them to drive every conversation: capture feedback, agree the scope, and only then write the PRD. VNPT approved the journeys first. The PRDs came after and mostly wrote up what we’d designed.

VNPT needed a credible proposition, engineering needed something it could build, and the agency needed room for a brand still taking shape. My job was to make the trade-offs visible, recommend the smallest change that kept the journey coherent, and stop decisions being reopened.

The home decision had to hold up in every situation, not just the two we’d argued about. So we designed four states. Usage, balance and navigation never move. What changes is the order below them, and one alert when the plan runs out.

Calls like the home screen couldn’t all come through me. So I wrote five experience principles, each one traced to a real screen. When requirements changed, and they kept changing, the team used them to make the next call on their own.

The hardest journey came first. Vietnamese eKYC, consent and customer data meant we couldn’t cut a single required step. So we made each one understandable, kept customers oriented, and gave every outcome a next step. VNPT approved the full journey, and engineering built the core flow from our specs.

The brand was still moving, so we built a system that could absorb change: tokens, behaviour rules and 40+ reusable components. DayUp stays bold in discovery and calm wherever customers share ID or pay.

Translation was the easy part. We reviewed tone, terms and imagery with VNPT, took cultural guidance from the agency, and benchmarked local carriers. The lucky-number finding showed up here too.

Handoff isn’t where quality is won. I set up a build review against the approved designs, with every gap logged in one ticket format. Once the design system landed, design issues fell from 22% to 10% of QA bugs. Most were small spacing and layout slips, the kind that creep in when many people build the same screens.

Then we let AI take the repetitive work on both sides. In August we got beta access to Figma agents and used them on new requirements: they turned our components into finished designs much faster, and gave the team breathing space. Engineering was already pushing for Cursor, so I proposed pairing it with Figma MCP. They could then generate UI straight from those same components, with fewer UI issues reaching review.

DayUp hasn’t launched yet, so there are no results to claim. Here’s what we delivered, and what we’ll measure once customers arrive.

Three lessons I’m taking with me.