XuChenChen Duoasa

How I got here

Long before I knew what UI or UX meant, I enjoyed drawing, taking things apart, building models, and putting them back together in my own way.

Back in middle school, I used to draw warships all over my English textbooks. What interested me wasn’t only their appearance. I liked how the hull, bridge, radar, mast, deck, and different functional systems formed something complicated but orderly—and actually worked as a whole.

That interest never really disappeared.

As a kid, I worked with paper, pens, models, and spare parts. Later, my tools became Sketch and Figma. Today, more and more of them are AI agents, code, and runtime environments.

The tools changed, but I’m still doing the same thing: starting with an idea and trying to make it real.

From design to running products

I studied computer science and technology at university, but my career gradually moved toward UI/UX and product design.

Over the years, design became my strongest discipline. I learned to work with interfaces, interaction, information architecture, visual systems, and product flows. More importantly, I learned that a product is not just a collection of screens. Every interface reflects decisions about what matters, how something should work, and what users need to understand.

I don’t describe myself as a software engineer. But computer science gave me a useful understanding of systems, data, state, APIs, and software behavior.

AI has brought these two parts of my background back together.

Instead of only describing how a product should work, I can now enter the real codebase, build a working version, use it, notice what feels wrong, and change it again.

I still use Figma when visual precision matters. It just no longer has to contain the whole product before implementation begins.

My process is usually:

Imagine → Build → Run → Observe → Judge → Iterate

The running product has become part of the design environment.

How I think about design

My taste has been influenced by Bauhaus, Braun, Sony, Apple, UniFi, and products that find beauty in function, structure, precision, and engineering.

But I don’t believe everything should be minimal.

Simple problems shouldn’t be made unnecessarily complicated. Genuinely complex problems shouldn’t be stripped down just to make an interface look clean.

Sometimes complexity needs to exist. The designer’s job is to understand it, organize it, and help people make sense of it.

That is why I care more about appropriateness than minimalism.

The best design doesn’t make people notice how clever the designer was. It makes the product feel like it was always supposed to work that way.

Nothing arbitrary. Everything with a reason.

What I’m building

QuotaView began as a small macOS utility for viewing Codex usage and quota. As I kept working on it, it became a broader exploration of human-agent interaction and agent observability: how people can understand an agent’s state, activity, progress, and limits without constantly supervising it.

DeepViewer explores the same subject from another direction. Built on DeepSeek Harness, it asks what a desktop workspace designed around AI agents could look like, and how files, code, previews, tools, generated results, agents, and human actions should live together.

Duoasa Design started as my UI/UX portfolio. It is gradually becoming my personal product hub—a place for the products I’m building, the experiments I’m running, and the ideas I’m still trying to understand.

I’m also turning lessons from these projects into reusable Agent Skills, including how-to-make-application and interactive-component-integration.

Instead of solving the same problem once, I want to see whether the judgment behind a solution can become a reusable rule—something an AI agent can understand without removing human taste from the process.

Where I am now

“AI-Native Product Designer” and “Design-led Builder” are probably the closest descriptions of what I do today.

Design is my strongest discipline. Computer science is my technical foundation. Product thinking helps me decide what should be built. AI expands how much of it I can build myself.

I’m less interested in collecting titles than in increasing the distance I can travel from an idea to a real product.

To define a problem, imagine a different answer, design it, build it, give it to real people, discover where I was wrong, and keep going.

In the end, I’m still the same person who likes making things.

I just use different materials now.