AI image generation has reshaped the role of the designer. In a world where nearly every result can look stunning on the surface, the true challenge lies not in making but in choosing—identifying meaningful directions, knowing when a variation is worth pursuing, and building an intuition for what works and why.
The abundance created by AI models has made process thinking more critical than ever. Good designers today must not only guide the model—they must understand and reflect on the path taken to reach an outcome.
This is where DUAI came in.
Role:
Product Designer, Solo Founder
Industry:
AI image generation
Duration:
2022 - 2025
Overview
DUAI was born from the desire to navigate generative AI more like a designer and less like a gambler. What if we could track every step, prompt, and decision we made along the way—like branches in a decision tree? Not only to reach better results, but to understand how we got there.
I introduced the concept of a generative decision tree, mapping each prompt, variation, and retry as a visual path. This concept evolved into a system that let users see their creative flow, compare outcomes, and highlight key moments—like turning points, deviations, and dead ends.

Challenges
AI design brought new challenges. The best designers weren’t the ones who made beautiful things (AI could do that). They were the ones who could choose wisely, detect nuance, and guide the AI with clarity and strategy.
Process invisibility — No native way to see or reflect on how a final image was reached.
Presentation difficulty — Freelancers and educators lacked tools to explain creative choices.
Prompt iteration chaos — Prompt history was flat, making it hard to track cause and effect.
Three needs shaped DUAI:
Tracking – Log prompt iterations and keep a visual record of choices.
Teaching – Present the process to clients, students, or peers.
Observing – Study what worked, what didn’t, and why.
We weren’t just generating pretty images. We were iterating. Learning. Building intuition in a world where everything looks good—but not everything is good.
My Approach
Rather than try to simplify or linearize the creative flow, I leaned into its complexity. I saw value in making the invisible visible—building tools that let designers see their own thinking laid out as a navigable map. My goal was to design a system that required minimal input but gave maximum insight: a visual decision tree that formed organically from user behavior. It needed to be fast, intuitive, and frictionless—something that felt like a natural extension of the generation process, not a burden.
I chose to operate within Discord (where most AI generation happened), and focused my interface design on two core objects: The generation Card and the Flow tree. With these, users could return to any point in their creative journey, analyze it, and continue with clarity.
Conclusion
DUAI wasn’t just a tool. It was a mindset shift—and a turning point in how I view my role as a designer.
It taught me how to:
Treat process as a product.
Give form to decision-making.
Build useful tools when the tools don’t exist yet.
Step into the role of a builder, not just a thinker.
Lead a project through uncertainty with vision and adaptability.
By navigating technical stacks, writing specs, managing scope, and refining every design decision under real-world constraints, I became not only more fluent in the languages of development and infrastructure—but more confident in pushing ideas from concept to execution.
It’s still one of my proudest explorations.