Creative Engineering Against the Same Design
Meng To opens Design Jam S1 with a number that sits on the table for the next two hours. A ten thousand dollar website. By the end of the session, he says, the room will know how to build one. Not as a pitch deck promise. As a craft problem. He is already tired of the default path.
I'm tired of seeing those little dots and those like stick figures generating HTML code. I'm tired of seeing always the same files, the same freaking design by everyone. Aren't you tired of that? Every design looks the same nowadays.
That is the room's starting line. Coding without design is no longer enough. The session is about moving off hobby output and into work that can hold a ten thousand dollar price.
From viral 3D to a paid craft
Meng walks through a scene he built in hours with current models. A boat. Sakura leaves. Water. Grass. Animations that read like a triple-A game from a few years ago. One post around that work got about half a million views in twenty-four hours, with thousands of likes and hundreds of thousands of page views on the site itself. He puts the money on the table without theater. A hosting company paid a thousand dollars for him to run the experiment on their product. Other sponsors message daily with offers in the one to five thousand dollar range for 3D experiments.
The point is not virality as a personality trait. It is that a designed experiment can become a commercial asset. Sponsorship, selling websites, and selling courses sit next to each other as paths. He was among the first designers selling courses online a decade ago around Xcode for designers. The same pattern holds now. Build something people have not seen, then decide how it earns.
He keeps returning to vocabulary. Steal the prompts. Say three.js. Say shaders. Say transparent PNG. Say Iconify. Say typography. Those words are how you steer. "Vocabulary are your prompts." Without that language, the agent fills the page with the same AI slot everyone else ships.
Creative engineering in the room
Meng names the practice creative engineering. Know the constraints of the stack. Know what three.js can do. Know what a shader button is. Know how layers separate from a background so parallax is possible. Then steer the agent into that space instead of accepting the first fade-in, fade-out everyone already knows.
He turns to the floor with a concrete exercise. How would you animate this text. Ideas come back from the room. Explodes. Snapping. Stagger. Randomize, like cryptography scrambling letters. He treats each answer as a prompt, not a joke. "We're here to practice your creative engineering." The same session pushes a related idea that sounds almost unserious on purpose. Stupid ideas that people have never seen before. Fireworks at the end of the boat. Two boats instead of one. Diablo, but Singaporean style. New is creative, and new is what travels.
I keep hearing builders ask how to make AI output look less generic. This Friday's answer was sharper than another mood board. Taste shows up as steering language, self-scoring, and the willingness to ship a strange idea before it is polished.
Steering, scoring, and what clients pay for
The practical half of the jam is a recipe more than a manifesto. Pull inspiration from Mobbin, Collect UI, Recent.Design, Behance, or a screen recording. Ask for five top sites, then screenshots. Mix sections across references. Plan nine or ten sections for a landing page. Then steer. Turn a static hero photo into three.js. Ask for separate transparent layers. Ask for ornaments. Ask the model to self-verify against screenshots until the page holds.
A technique he has been using: make the agent score every element out of ten, then keep working until it hits eight or nine depending on token budget. Left alone, he says, it often starts around four or five. Treat the model like a design agency. Agencies ask for more budget when quality goes up. So do tokens.
That is how a client sees ten thousand dollars of work. Not only the final page. The brand guide, typography, motion, logo exploration, and interactive pieces that used to take weeks. One prompt can now produce a large stretch of that pack. The agent does the labor. The builder still has to show the process and taste the result.
In Q&A he is blunt about aesthetics. "Aesthetics for me is function." A hospital with only white walls works against healing. The same applies to product UI. Function without taste is still incomplete. He has not touched code line by line in about ten years. What remains is testing, tasting, libraries, prompts, and steering until the thing in the user's hands is right.
What to watch next
The interesting thread from Design Jam S1 is not another list of tools. It is whether builders in Singapore will treat creative engineering as a priced craft, with sponsorship and high-ticket sites as normal outcomes, or keep shipping the same slot everyone already scrolls past. Meng left the room with prompts, a scoring habit, and a bias toward ideas that look a little stupid until they get half a million views. The next Friday is for whoever tries that on a real brief.
SQ Collective hosts Coworking Fridays for founders, operators, and AI builders working through real product questions in Singapore.
Join an upcoming Coworking Friday: https://lu.ma/ai-labs
Explore SQ Collective: https://www.sq-collective.com
Michael
Comments