The useful thing about hunger is that it makes abstraction difficult. At three in the morning, after a work call, Andy Lin did not want a market map. He wanted hot soup. The electrical engineer had spent years in the semiconductor business, where precision is assumed and systems are designed to run. Yet the food system around him had a bedtime. Nearby options had narrowed to a Denny’s. Somewhere between the craving and the compromise, a question took hold: why could a warm bowl not be available whenever someone needed it?
It was 2015. Lin began looking for an automated answer and found examples of old Japanese noodle vending machines, but not the kind of fresh, connected cooking system he imagined for the United States. So he started building. About six months later he had a prototype. It worked, in the broad and generous way that prototypes work. It was also roughly the size of a minivan.
That ungainly box is the best introduction to Lin. Today, Yo-Kai Express machines are compact enough for airports, campuses, hotels, factories and office buildings. They prepare hot meals in well under two minutes and operate around the clock. But the polished machine can hide the stubborn physical work that came first: heating, timing, packaging, restocking, cleaning, menu design and the thousands of small accommodations required when software meets soup.
An engineer changes his material
Lin studied bio-industrial mechatronics at National Taiwan University, then earned a master’s degree in electrical engineering at UC Irvine. He spent about 12 years in semiconductors, including engineering and technical-sales work at ASE Group. The résumé makes his move into ramen look like a sharp turn. In practice, he changed the material more than the method.
A semiconductor process breaks a complex outcome into controlled steps. Yo-Kai Express does much the same with a meal. Temperature has to rise on schedule. Ingredients have to survive storage and emerge with texture intact. Containers have to tolerate a journey from deep cold to cooking heat. Inventory has to appear in the right machine before someone taps the screen. A bowl is the final output of a system most customers never see.
Lin had already tried another piece of the food problem. In 2013 he began Sholics, a last-mile grocery-delivery app. Delivery taught one lesson about moving food; the late-night soup problem suggested another. Instead of sending a driver from a restaurant to a customer, why not put the cooking system where demand already exists?
Since we are a startup company, we all wear multiple hats, from fundraising, marketing, sales, and design; we all do it collaboratively.Andy Lin
The word “collaboratively” matters in a hardware-and-food company. A software bug can be patched invisibly. A bad bowl arrives in public. The product requires mechanical engineers, food scientists, chefs, operators and salespeople to agree on the same minute of the customer’s day. Lin’s job has expanded with that system. Founder, engineer, fundraiser and evangelist are not separate shifts. They overlap.
The machine is not the business
The obvious label for Yo-Kai Express is “ramen vending machine.” Lin has spent years pushing past it. A vending machine is a container with inventory. His larger pitch is a food platform: restaurant recipes translated into repeatable cooking processes, then distributed to places where a conventional restaurant is expensive or impractical.
This is why the company has worked with established operators and brands, including Aramark, Compass Group, Sodexo and Japanese ramen company Ippudo. A restaurant usually scales by signing a lease, building a kitchen, hiring a crew and recreating the experience block by block. Lin’s model proposes another route. Put the recipe into a controlled system, move the system into existing traffic and let the dining room disappear.
The name carries the idea. A yo-kai is a spirit or mythical creature in East Asian folklore, able to appear at unexpected times and places. Lin chose it for food that could do the same. The playful name disguises a serious distribution thesis: a chef’s reach should not end at the edge of the dining room.
By 2021, the machines could hold more than 20 menu choices and 88 meals per fill. The menu had moved beyond Japanese ramen into udon, pho, Taiwanese beef noodle soup, rice dishes and desserts. By 2023, the company was reporting more than 150 sites across the United States, South Korea, Japan, Taiwan, Singapore and the United Kingdom. The point was no longer that a robot could make one bowl. It was that the same distribution logic could hold many kinds of comfort food.
That variety creates a less visible engineering problem: deciding what belongs in each machine. A kiosk near an airport gate does not behave like one in an office or on a university campus. Demand changes with the hour, the weather, local events and the rhythm of the building around it. Yo-Kai Express has described using sales and location data to adjust replenishment, with the longer-term goal of bringing traffic, event and weather signals into the forecast. The robot may cook the meal, but an empty slot still defeats the system.
This is also where Lin’s automation argument becomes more practical than the familiar robots-versus-workers debate. The cooking point can run without a traditional kitchen crew, while production, stocking and maintenance remain human jobs. Automation moves the work across the network. It trades a cook waiting at every counter for a coordinated operation behind many counters. Whether that trade works depends on details customers rarely notice - a bowl available at the right site, the right hour and in the right quantity.
A decade measured in prototypes
Expansion also exposed a cultural wrinkle. In Japan, automated ordering needs little explanation. In the United States, customers often have to be taught that the machine is cooking rather than dropping a shelf-stable cup. Lin’s job is partly translation: between chef and engineer, between machine and customer, and between markets with different ideas about what automation means.
His answer has been to keep widening the demonstration. Yo-Kai Express added a boba system. Lin began talking about edge AI and virtual assistants that could respond to customers at an unattended location. “Unattended doesn’t mean unattended,” he said while describing the problem. The phrase captures a tension inside automation. Removing an on-site crew can extend service hours, but the person at the machine still wants acknowledgment, help and a sense that someone designed the moment for them.
At CES 2026, Lin appeared on a panel about robot stores and next-generation food kiosks. At NRF in New York, he discussed the company’s move from vending systems toward robot chefs. The ambition now includes a machine that can interpret a request, translate it into machine instructions, use vision to check the result and adjust the cooking. A customer asking for a medium-rare steak with more spice is a more complex interaction than tapping a ramen button. It is also a clear destination for Lin’s decade of work: customization without requiring a staffed kitchen at every location.
We want to build not just vending machines, but a cross-border food platform.Andy Lin
The bowl remains the test
Food technology invites theatrical language. Robots cook. AI recommends. A restaurant fits into a box. Lin’s story is more interesting when reduced to ordinary standards. Is the meal hot? Does it taste as intended? Is it available when the customer arrives? Can the operator keep the machine stocked? Does the economics work without the illusion of a restaurant around it?
Those questions are less glamorous than a robot chef and more consequential. They also explain the long march from minivan-sized prototype to connected kiosk. Lin has not been building a single dramatic invention. He has been removing one operational excuse after another: the kitchen is closed, the location is too small, the chef is too far away, the line is too long, the hour is too late.
There is a personal consistency underneath the expansion. Lin still talks about noodle soup as comfort food. The company can add boba, steak, computer vision and dynamic inventory, but its founding observation remains intimate. People do not experience a food system. They experience being hungry at a particular hour in a particular place.
A decade ago, Lin met that moment with an engineer’s refusal to accept the available options. His first answer occupied the space of a vehicle. The machines became smaller. The idea became larger. If it works, the autonomous restaurant will not feel like a futuristic destination. It will feel like something much more practical: the warm meal that happens to be there.