At lunchtime, Kevin Yang had a question to answer. So did everybody else. Where was the order? At EAT Club, the food delivery company he co-founded, the support requests arrived in a rush. Answering one meant checking an order, finding out what was happening with the restaurant and driver, and deciding what to tell the person waiting to eat. Repeat that across a crowded lunch hour and a short question becomes a considerable occupation.
Years later, Yang would still be talking about that hour. By then, he was leading AI at Front, and the experience had become a concrete example of the work he wanted software to handle. The continuity is appealing. A founder starts by trying to get lunch to a desk and ends up working on how companies coordinate their response to a customer. The sandwich has left the conversation. Its whereabouts remain relevant.
Between those two jobs came Idiomatic, the customer feedback company he helped build and led before its acquisition by Front in November 2024. His route through the three businesses offers a useful way to look at his career: follow the unanswered question. Each enterprise gave him a different place to stand while considering it.
A lesson from Mumbai, an argument in California
The initial question was simpler: why was office lunch so disappointing when good restaurants existed nearby? Yang and Rodrigo Santibanez had experienced the vending machines and uninspiring nearby options themselves. At Stanford Business School, they studied Mumbai’s dabbawalas, the delivery network carrying meals between homes and offices. Yang recalled a lesson that emphasized how particular the system was to India. He and Santibanez saw something they wanted to try in America.
They started EAT Club in 2010. The early service offered rotating restaurant selections ordered online for delivery to office workers. By 2013, Yang was describing an alternative to leaving the office to hunt for lunch: choose in the morning, have a meal arrive at the desk. Prices then started at $8.95. That historical detail captures the original proposition neatly. Convenience needed to be ordinary enough to become a habit.
The founders divided the work in a way that made the operation tangible. Santibanez recruited restaurants and customers and organized delivery. Yang wrote the first website and handled customer service. Their working space moved from Yang’s dorm to Santibanez’s second bedroom. There are glamorous ways to describe starting a company. Converting a spare bedroom into an office has the advantage of being measurable in square feet.
That division put Yang close to both the machinery and its failures. He could build the ordering experience and hear what happened when the experience disappointed someone. In a business with a daily deadline, the feedback loop was unusually short. A customer waiting for lunch had little reason to save the complaint for a quarterly survey.
“Where’s my order?”
- 01Check the order
- 02Check restaurant & driver
- 03Decide the next step
- 04Update the customer
The menu had a memory
By February 2015, EAT Club had delivered more than two million meals and accumulated roughly 375,000 reviews of individual dishes. Nearly half its employees used its Looker analytics system daily, including people buying ingredients, preparing food and delivering it. Yang described a business that could connect a returning customer with an order and a response. A conventional restaurant might watch a diner leave without learning any of those things.
One example concerned vegetarian meals. Yang estimated that about eight percent of EAT Club’s customers were vegetarian, yet vegetarian dishes accounted for 23 percent of sales. Customers who also ate meat sometimes wanted a vegetarian lunch. Planning around a fixed label would have produced the wrong menu. The numbers made room for people to behave like people, with preferences that changed from one meal to the next.
This is a revealing episode in Yang’s working life because the data had an immediate destination: somebody needed to decide what to cook. Counting preferences was connected to buying food, arranging menus and avoiding waste. The purpose was practical. A chart that helped a buyer place an order had earned its place in the kitchen.
Appetites don’t fit neatly into boxes.
Lunch grows up
EAT Club continued to expand. In April 2013 it announced a $5 million funding round, with plans to move beyond the Bay Area. The service could also work with employers subsidizing meals for their staff. That arrangement brought a second customer into the picture: the business organizing lunch alongside the individual choosing what to eat. The product had to accommodate both.
In 2017, a $30 million Series C led by Sodexo backed further expansion. EAT Club described serving tens of thousands of individual meals daily to offices in the Bay Area and Los Angeles. The central arrangement remained recognizable: employees could choose for themselves while their employer ran an organized lunch program. Individual choice and coordinated delivery had to coexist.
By 2018, the company was reporting more than ten million meals delivered since launch. Those figures describe the company’s development across years and a wider team. For Yang’s personal story, the more consequential inheritance was the work he had done close to the customer. He had spent enough time answering questions to recognize how much useful information arrived disguised as another support request.
A second company in the incoming mail
Idiomatic approached that information as a product. Its pitch was to organize the problems customers described in support tickets, survey comments and app reviews, turning written feedback into something a business could count and use. Yang worked alongside co-founder Christopher Martinez. The idea connected customer service with decisions about what a company should improve.
In a 2022 article they wrote together, Yang and Martinez described a familiar imbalance: the loudest complaints can command attention while recurring problems affecting quieter customers remain unresolved. Collecting more feedback does little by itself to correct that imbalance. Someone still has to understand the pattern. Their argument was for a fuller, timely picture of what customers were saying, rather than a business steered by whichever complaint arrived most dramatically.
The distinction matters to a founder who has answered the phone. One unhappy customer deserves a response. A recurring problem deserves attention beyond the response. Idiomatic’s work occupied the space between those obligations. Its eventual customers included HubSpot, Instacart and Pinterest. Different businesses, different conversations, and the same practical need to learn from them.
In November 2024, Yang announced that Front was acquiring Idiomatic almost nine years after its beginnings at his kitchen table. He thanked colleagues, customers and investors for committing their careers, businesses and money. Martinez described the work as the longest, most rewarding and most challenging undertaking of his professional life. Those acknowledgments give the transaction a human scale: years spent building something together before handing it a new address.
The inbox gets a new address
Front brought Yang in to lead AI strategy and products. The acquisition also brought the Idiomatic team into work on Front’s AI roadmap. Customer conversations were already central to Front’s service platform. Yang arrived with years of experience trying to extract useful meaning from those conversations and a much older memory of what it felt like to handle them while running a delivery business.
He described those early years as teaching him “an enduring respect for customer service professionals.” That is a useful sentence to carry into a job designing automation. People handling support hold together the parts of a business that customers encounter when something goes wrong. Their work includes investigation, explanation and the occasional awkward decision. Making the reply quicker addresses only part of the assignment.
In April 2026, Front used Yang’s EAT Club experience to explain new AI features. The company presented Autopilot Playbooks for processes across systems, Autopilot Resolve for customers completing requests themselves, and Copilot for assisting staff. The old lunchtime question provided the example: an answer depended on information and actions spread across an operation. Yang’s earlier job had supplied a problem specific enough to test a later product idea against.
“Making customers happy is a team sport.”Kevin Yang
Physics, a frying pan, and the next thing
The résumé behind these companies wanders more than the sequence of startup names suggests. Yang grew up in Kaohsiung, Taiwan, and suburban Chicago. He studied physics at Harvard, graduating magna cum laude, and earned an MBA at Stanford. His personal biography lists work ranging from teaching physics and translating Classical Chinese to consulting, venture capital and developing online games. It is a career with several changes of vocabulary.
Cooking provides a smaller recurring thread. Yang’s personal page ends with the claim, “I make a really excellent steak.” A later professional biography describes him cooking for his wife and three daughters. It also identifies his favorite quotation as Anna’s instruction in Frozen II to do the next right thing. Family dinner and a Disney lyric give the biography some welcome texture beside the company names.
His public work in 2026 continued to return to customer service. An August podcast appearance with Greg Kihlström considered speed and coordination in AI. In September, Front quoted Yang emphasizing the continuing value of human connection in service. The ambition is consistent with the problems he has chosen to work on: help a business understand the person asking for assistance and equip its people to respond.
Follow Yang’s story from the dorm room to Front and the customer remains remarkably close. At one end is a founder answering calls while building a website. At the other is an AI product leader thinking about how work moves between people and systems. The intervening years have changed the tools and the scale. At noon, somebody still wants a useful answer before the meeting starts.