Prototyping in Code:
A Monthly AI Beta Loop
Static mockups were slowing learning on a product that shipped every month. Prototyping in code, alongside engineering, closed that gap.
PlayStation · PS App mobile betas
2026
Set the cadence with my engineering and product partners
My designers, an engineer, and a product manager
iOS and Android
The situation
Mobile released monthly while console ran on longer cycles. Handing engineering static files and waiting for a build slowed learning in the one place the team could learn fastest.
What I decided
Prototype in code, in parallel. Instead of handing off static files, designers used AI coding tools (Codex and Claude Code) to prototype while engineering built in React and React Native.
Define "ready" together. Design and engineering jointly checked whether the experience met the bar for an MVP before anything shipped.
Get real feedback early. Players could opt in on the mobile app to receive small releases and share feedback.
Monthly
Release Cycle
~3 weeks
Prototype to rollout
~1 week
Review per cycle
How the team carried it
Each month followed the same loop: a week of planning and alignment, prototyping alongside engineering, an MVP check, rollout after about three weeks, then roughly a week to review impact.
What changed
The team established a repeatable monthly rhythm for learning from real users. Measured results from the cycles were still pending when I moved on.
Stack
Codex · Claude Code · React · React Native · Figma
“Whether collaborating on strategic work or exploring ideas beyond our day-to-day responsibilities, she approached each opportunity with curiosity, purpose, and a genuine commitment to creating meaningful experiences.”
Porfirio Moreno Ortega
Product Designer at PlayStation
What it taught me
AI is most useful when it shortens the distance between an idea and real feedback. The tools mattered less than the habit of testing small and often.