Leaflyze: Dreaming of a day when robots take up sewing machines

Leaflyze, the future of sewing changed by a crazy entrepreneur

“No, you still make dolls by hand?”

When you invest, you sometimes come across scenes that are hard to believe.
We live in an era where AI makes movies and robots serve customers at restaurants.
In garment factories, many people still sew all day long.

Seriously, by hand.

Among the 'food, clothing, and shelter', 'clothes' are distributed, resold, and washed, etc.
This is a field where startups and unicorns are constantly emerging.

But how clothes are actually made remains the same as before.

Design → Pattern making → Cutting → Sewing.
These four steps are still done manually, one by one.
Create a pattern with a designer's intuition, cut the fabric with scissors, and sew it with a sewing machine.

The results of that creation are the clothes we wear every day, the dolls we hug, and the bags we carry around.

Dolls made by Leaflyze

“Can’t we automate this?”

The question was simple, but the beginning was difficult. The sewing industry has been a blind spot for automation. Fabrics are crumpled, folded, stretched, and even dried. The physical properties of each material are different, and the tension is different when pulled horizontally, vertically, and diagonally. For a machine to handle this, a truly enormous amount of data is needed. Leaflyze sees that distant future and is accumulating data one by one from now on.

In February 25, I visited a garment factory on the outskirts of Jakarta, Indonesia, with CEO Jong-ho Park. After a three-hour drive, hundreds of workers were crowded together and operating sewing machines. They were still doing it by hand. When I asked if there was any automated equipment, the answer I got was always the same. “You’re automating this? ” I could tell in just five minutes on site that this was not a joke, but serious.

The bigger problem is that factories move around to meet labor costs. From China to the Philippines, Indonesia, Vietnam, and Bangladesh. But now there is no place to move anymore. It takes one month to procure raw materials, and 1.5 months to transport them back to the US after production. It is time to break this inefficient cycle that takes two and a half months. And the team that started this challenge is Leaflyze.

Sewing Factory, Jakarta, Indonesia (2025.02)

Can machines replace the work that people do?

Leaflyze focuses on three things: pattern making, cutting and sewing.

The goal is to automate these three steps with AI and robots. The designer takes a 2D image and the AI converts it into a 3D design, which automatically generates a pattern that the robot can work with. The important thing here is that this pattern is not designed for humans, but for machines to actually move .

The key to making this technology possible is simulation. There are two things to it.

•

Forward simulation is predicting “what 3D shape will be formed when this is folded?” based on a given pattern. In simple terms, it is like seeing what kind of doll will be formed when the paper is folded according to the design.

•

On the other hand, the reverse operation is much more difficult. For example, if a request comes in like “Please make this doll’s ears 2mm pointier,” the machine has to figure out what pattern to recreate from scratch to match that request.

This is not simply a matter of changing the appearance, but rather a matter of re-drawing the entire internal diagram if the appearance is changed . This is something that is difficult for people to do intuitively, and much more difficult for machines.

The person who makes it possible is CTO Seunghwan Lee. He is an expert in 3D and physics simulation, physics-based motion control, and data-based motion learning and generation , having received a Ph.D. in computer science from Seoul National University and worked as a postdoctoral researcher at Stanford. In simple terms, he is a person who calculates how machines should move naturally in the real world using mathematics and physics. This is because precise modeling and learning are necessary for robots to handle flexible and complexly responsive materials like fabric.

And in order for this technology to be implemented in actual robot equipment, the process of precisely connecting and feeding the generated data to the machine is important. This part is handled by CPO Jongmin Park, another co-founder. He received his bachelor's and master's degrees in computer science from KAIST and has experience leading the backend development of the collaborative robot platform at Tweeny. He is in charge of making the robot pattern data created by Leaflyze operate accurately in actual equipment.

CPO Jongmin Park (left) & CTO Seunghwan Lee (right)

Why is this founder called a 'crazy person'?

At Basis, there's a question that we always ask when investing: "Why is this founder crazy?"

CEO Jong-ho Park is the founder who answers that question perfectly.

People who are attracted to 'doing something completely new' rather than 'doing something that is just a little bit better'.

During his time at Suarab, when most competitors were focusing on accumulating more data and increasing model accuracy by 1-2%, he persisted for five years, saying that he would create a model that could be used in the field immediately with only a small amount of data , and eventually achieved results.

Upstage had been working on world-class OCR projects, but he felt a greater thirst. Rather than increasing the accuracy by 1% from 95%, his direction was to create a completely new board. So he set the problem himself and chose the path of entrepreneurship to change the board himself.

Of course, the process of starting a business was not easy. The items were folded several times and the team was scattered.
But he didn't stop there. He met customers in person, toured factories, and found real problems.

Leaflyze Park Jong-ho, Co-founder & CEO

A team that can make your crazy dreams come true

The world won’t change with just the founder’s ambition. Leaflyze has the team to make that dream come true.

How to define AI learning data for sewing automation, what properties should be reflected in the simulation, how to reduce the error between the simulator and the real robot. These are not problems that can be solved simply by being good at coding.

This team knows how to do it. And they're doing it.

Leaflyze members

The history of the sewing industry will be rewritten

Leaflyze is already working on real projects with large ODMs like Aurora World and long-tail goods brands. What's next? We will challenge the automation of almost all the sewing products we wear and use - from clothing to bags and shoes.

In the past, factories moved with labor costs and people . In the future, factories will move with markets and data . That's because of the changes that Leaflyze is making.

Two years ago, CEO Jong-Ho Park said to me the first day we met.

“I want to make money. But not in a hole-in-the-wall kind of way. I want to solve really big problems and create trillions of dollars of value.”

Now he is solving that problem.
So BASS Ventures invested in that crazy dream.

From dolls to clothing, shoes, and bags, the future of sewing rewritten by Leaflyze

Subscribe to 'BASS Ventures'

Welcome to BASS Ventures, the mecca for entrepreneurs with crazy dreams!
By subscribing to the site, you will be the first to receive the latest updates, including new posts, via notifications and email.
Join SlashPage and subscribe to 'BASS Ventures'!