How Far Can a Product Designer Build with Codex?

AI summary focused on the core points of the original article.

Read original

1. Overview and Background

Author: Jude Park, a product designer at Grab.
Project: A shuttle bus operations system for Korean Presbyterian Church in Singapore.
Core question: In an era where generative AI (OpenAI Codex) lowers the barrier to code implementation, how much of the product’s full development lifecycle (SDLC) can a designer directly own?

2. Step-by-step Build Process and Feedback (Build Loops)

Step 1: Build a web MVP (Next.js, Prisma, LINE LIFF)
- To validate quickly, the team first defined the plan and API spec (Swagger) before starting development.
- Implemented basic features such as route lookup, QR boarding check-in, and admin notifications.
- Feedback: The functions worked, but access through a messaging app created UX limitations.
Step 2: Expand into native apps (Rider App) (SwiftUI / Kotlin)
- Used the validated web API as a blueprint to develop native iOS (SwiftUI) and Android (Jetpack Compose) apps.
- Applied each platform’s latest design language (Liquid Glass, Material 3 Expressive).
- Feedback: Discovered a gap in empty stops where passengers do not scan a QR code, making it impossible to track the shuttle’s actual location / whether it is in operation.
Step 3: Build a Driver App to establish a 'single source of truth'
- To get a reliable location signal, the team separately built a driver app and changed the structure so the bus itself publishes location data.
Special note (mobile voice coding): While staying at a hospital in Korea to care for his father, he used ChatGPT mobile’s voice interface connected to a Mac mini environment in Singapore to issue commands to Codex and carry out remote development and testing.

3. Key Takeaways and Lessons

An expanded designer role: Goes beyond simple screen design or handoff to lead problem definition ➔ data/API modeling ➔ native app implementation ➔ app store launch and promotional video production.
The essence of AI use: AI is a powerful partner that reduces implementation distance, but deciding “what the real problem is” and setting the product direction ultimately depends on the designer/planner’s product judgment.

XPLORED View

This piece is a great example of how far a designer can go: it makes you feel, “So designers can actually build this much.”

Because it starts small and pushes all the way through to real deployment, it’s useful for getting a feel for how AI tools fit into practical work.

It’s especially recommended for anyone who wants to look at maker power and product judgment together.

Read original

This article is summarized and interpreted by XPLORED based on publicly available original sources and related materials. Rights to the original source and materials included in it belong to their respective rights holders. If a rights holder requests a correction or removal, we will review and respond promptly.

Get the week's signal by email, every Saturday.

A short read on the tools and articles added that week.