The first thousand customers had done exactly what a young company hopes customers will do. They had ordered early. Duy Huynh’s team promised to put them first. Then, in the confusion of a launch that attracted more than 3,000 pre-orders, the company sent products to the later buyers instead. The queue had learned to walk backwards.
Huynh has recalled the episode as embarrassing, with customer service overwhelmed by complaints. The lesson he drew was about details: anticipate what might go wrong, and get help when the task exceeds what one person can manage. It is an appropriately unglamorous opening for a career that has included a robot with a face, a mobile marketplace and a motorized desk.
The delivery mistake belongs near the beginning because it makes the rest of his story easier to understand. Making something people want creates a fresh set of obligations. The people who believed first are entitled to remember what you promised.
A factory before the future
Huynh came to the United States from Vietnam at 18. His education moved quickly: he completed an undergraduate degree at the University of Wisconsin in 16 months and entered doctoral study at the University of Maryland at 19. He subsequently worked as a software architect at IBM.
By 2007, he had founded VIFAH, a manufacturing business supplying consumer products to major retailers. That earlier experience matters to the later furniture company. A desk occupies two worlds at once: the drawing where it is designed, and the less forgiving world of components, production schedules and boxes.
In March 2011, he founded Taap.it, a mobile marketplace for local buying and selling. His ambition was to connect what people could discover online with what shops actually had nearby. By November, he said about 5,000 New York businesses had signed up and more than 70,000 items were posted.
He was also becoming attached to New York. Asked about the city, he joked that rent control might explain why he stayed. Beneath the joke was affection for a place where he enjoyed both working and living. His earlier moves had taken him through Wisconsin, Maryland and North Carolina. New York had persuaded him to linger.
The customer wanted Alfred
Robotbase followed in 2014. By January 2015, Huynh was presenting a personal robot with an ambitious job description: move through a home, recognize people, respond to requests and connect with other devices. It was the kind of product that asked a buyer to imagine a different domestic routine.
The prototype was called Maya. Huynh pictured a category between industrial machinery and toys, with a machine useful enough to become an ordinary presence. Early demonstrations included photography, household automation and assistance with schedules. These were promises being developed around a prototype, rather than evidence that every proposed feature had become an everyday service.
He expected potential backers to ask difficult technical questions. Instead, a prominent concern was whether the face could be changed. Could it be male? Could it resemble Alfred from Batman? For all the effort involved in making a machine recognize a person, the person wanted some say in whom the machine resembled.
There was substance behind the theatrical possibilities. Huynh described a robot using a mobile base and depth camera, with wireless connections, sensors and software interfaces for developers. A touch interface was meant to make it easier to use. The proposition involved both an object and an invitation to other people to build applications for it.
That invitation was central to his thinking. We want to be a platform.
A robot could gain capabilities through software. The difficulty was persuading someone to buy the initial machine while that larger possibility was still taking shape.
Huynh also described the coordination problem with unusual plainness. Robotics meant bringing hardware, electronics, software, mechanical design, user experience, manufacturing and distribution together. Each discipline could advance on its own and still leave the whole thing waiting. A business with seven moving parts has many opportunities to discover an eighth.
A desk with a business attached
Later, Huynh acknowledged that the team had concentrated on things it liked making, with too little understanding of marketing and demand. Maya’s technical appeal had run ahead of the business case. The entry costs were high for a young company, and office customers had not demonstrated the demand the team needed.
The move toward SmartDesk and ErgoChair gave Autonomous a more familiar place in the working day. A customer understood what a desk was for before encountering its engineering. That familiarity shortened one part of the conversation, even as manufacturing and fulfillment remained demanding.
Huynh has described the company’s early setting as a cramped New York apartment, occupied by robotics engineers and tinkerers. In 2015, crowdfunding put their ideas before early adopters. He remembered no single moment of revelation. Working with those buyers helped the team understand preferences it had not known in advance.
“All startups make dumb mistakes.”
Duy Huynh
By June 2017, he was reporting annual sales above $10 million, roughly 100,000 units shipped and $200,000 raised. These were his figures at that point in the company’s life. They describe an early business with substantial sales relative to the seed funding he discussed, rather than a measure of its present size.
June 2017 snapshot • Founder-reported figures
The product range also included a $19 cardboard standing desk. Set beside the robot, that object offers a pleasingly wide definition of technological ambition. The price of entry could be lowered all the way to folded cardboard. A useful idea did not require every possible electronic component.
By March 2020, Autonomous was describing its business around direct sales and control of production. Its public account emphasized the physical objects people touched every day. In that telling, the earlier ambition survived through a change of emphasis: more attention to ordinary tools and to what buyers would use.
A company cannot be debugged alone
A 2017 conversation about standing desks shows Huynh doing the less ceremonial work of being a founder. He identified himself, answered a shipping question and invited suggestions. He described five distribution centers, including New York, California, Toronto, Amsterdam and Shenzhen, as part of an effort to reduce shipping costs.
Such exchanges have a different texture from a launch presentation. Someone wants to know whether an order can reach them affordably. The reply has to meet that question at its own scale. A vision for the office of the future still requires an answer about the delivery destination.
In his later account of product development, Huynh described asking staff and customers about ideas, making prototypes, sending them for trials and revising them. He also described himself as a perfectionist and advised other founders to stop micromanaging. Colleagues needed room to think and develop their own work.
That advice carries a useful personal tension. The founder who cares about details must decide which details require his intervention. Following every task can consume the time needed to judge the larger direction. Delegation asks a perfectionist to tolerate the possibility that another person’s method will be different.
His working day, as he described it in 2021, had boundaries as well as tasks: an early alarm, taking his children to school, periods reserved for concentrated work and evenings with family. For someone selling tools for the working day, there is something revealing in the effort to give his own day an ending.
More room, more responsibility
The company eventually expanded the size of the object itself. WorkPod offered an entire backyard workspace. One published configuration had 102 square feet of floor area and arrived in three large crates. The office had become something the customer could put outside the house, complete with its own enclosing walls.
The progression has an almost comic physical logic: first the machine that might help around the room, then the furniture, then the room. Each step brings another layer of practical questions. Where will it go? How will it arrive? Who will assemble it? The buyer’s circumstances keep enlarging the design brief.
Huynh’s wider business record also contains a serious question of financial disclosure. A March 2025 Washington consent order against CONST LLC, doing business as MyConstant, identified him as its sole known owner and managing member. It described an undisclosed loan of more than $4 million to Autonomous, whose founder and CEO it also identified as Huynh.
The order concerned unregistered securities and failures to disclose material information. CONST LLC neither admitted nor denied the findings and conclusions. The respondent was the company; the order was not a criminal conviction of Huynh. These distinctions matter, as does the underlying question of what people financing a business are told.
The useful thing
Across Huynh’s career, the interesting decisions happen where an idea meets somebody else’s expectations. A buyer wants a different robot face. A shop needs its stock discovered. An early supporter expects the promised place in a queue. Each person supplies information the original design could not contain.
A desk makes that relationship unusually visible. It has to occupy a real room, fit the work being done and survive the journey there. Its maker can attach software and imagine new capabilities, but the person opening the box still gets a vote. Huynh’s education has involved learning how much that vote can change.