
OVERVIEW
THE PROBLEM
Retailers’ content lived in the CMS, with no easy way to turn it into app pages. They published without a real preview and depended on tech teams for simple changes.
WHAT I DID
End to end, reporting directly to the CEO: architecture, scope, every flow and state, a staged launch and QA with engineering.
IMPACT
more weekly publications in 3 weeks of beta (from 11 to 47)
My process
01. Research
I benchmarked Tapcart, Webflow, Firebase and Framer, and analyzed VTEX, the CMS where retailers’ content lives: great for storing content, not for designing a storefront.

02. Define
existing apps analyzed to scope the first version
App Studio sits next to branding, notifications and dashboards, so I designed each screen to solve today’s problems without limiting tomorrow’s.

I validated technical feasibility with developers and split the scope into sprints.
03. Design
Retailers drag components onto the canvas and set their content, look and destination. I designed every path and state, from errors to saves, for very different users: with or without AI, first app or existing one.
All of it runs on the design system I also built from scratch.
04. Test
In beta, I tracked what people actually did with Google Analytics 4.
15.6 min avg. engaged time per visitor · 407 component edits · 1 user published, in week one.
05. Deliver
Three stages: a preview with the retailer’s real content, then visual editing, then publishing with drafts, versions and rollback.
Designing in code
Beyond handoff and design QA, I vibe code parts of the product with Claude Code, building changes on branches and merging them.
A Minimum Lovable Product, not just an MVP
Why: retailers had to want to come back, not just be able to use it. Trade-off: a smaller first scope, polished instead of broad.
A visual layer on top of the CMS
Why: content stays where retailers already manage it: nothing new to maintain or migrate. Trade-off: the editor depends on how each retailer structures their content.
Reuse what already exists
Why: brand colors, components and the app’s own design system make the preview exactly what customers will see. Trade-off: less freedom to design outside the system.
Outcome
51%
of all engaged time on the platform, in week one
1,000+
component edits by clients in just 3 weeks
3
clients published their app for the first time in a single week (vs. 1 in week one)
It became the platform’s most-used module from day one, with 22 clients exploring it in a single week. Next, I’m looking into why deletions almost matched additions in the busiest week.
I keep iterating with users and my developers (my partners in crime). One of the most rewarding projects of my career.
In constant
motion
© 2026