The locksmith arrived in an unmarked van. Greg Marsh was away, his wife was locked outside their New York apartment, and the eventual bill was $300 for drilling the lock. It was 2011, and the couple had only recently moved to the city so Marsh could attend Columbia Business School. Most people would have kept the receipt as an expensive souvenir. Marsh kept the question: why did getting back through your own front door feel so uncertain?
He knew nothing about locksmithing. That turned out to be useful. Research showed him a large, fragmented trade with very little technology and a customer experience often defined by urgency. A locked-out person cannot browse at leisure. The imbalance between a calm service provider and a stranded customer was practically built into the transaction.
Marsh imagined a machine that could scan a key, remember its geometry and make another copy when the original disappeared. The object would remain pleasantly old fashioned. Everything around it - computer vision, secure digital storage, robotics and retail distribution - could be rebuilt.
“I actually viewed dropping out of business school as the least risky option.”Greg Marsh, on committing to KeyMe
First cutThe least risky way out
Leaving Columbia sounds like the cinematic gamble in the story, but Marsh's reasoning was almost actuarial. Classmates who divided their attention between school and a startup gave the company too little. If they waited until graduation, failed, and then needed a job, the ordinary recruiting season had already passed. He could return to school if KeyMe failed. Giving the idea a full attempt, in his view, reduced the muddle.
His earlier career had prepared him in pieces. Right out of college, he worked in finance on a six-person team managing a $1 billion portfolio. The numbers interested him, but operating did more. He left to become the first employee at Aperia Technologies, a Bay Area hardware startup working on tire inflation for commercial trucks. There he saw product building and fundraising without the flattering distance of a spreadsheet.
At Columbia, he went looking for people who could turn the key idea into machinery. His tactic was gloriously literal: find engineering professors whose work seemed relevant, look up their office hours and appear. Business students did not commonly do this. Professors noticed. Engineers joined. In a university lab, the early group combined cameras, software and an off-the-shelf CNC milling machine into a proof of concept.
Prototypes are where elegant ideas acquire smoke. During one early session, a laser used to cut plastic started a small fire. The team did not burn down the lab, Marsh later recalled, but Columbia politely suggested they continue elsewhere. It is hard to devise a tidier metaphor for a hardware startup: make something physical, discover the physical world has objections, then carry the pieces to another room.
The corner testCustomers without a script
The better omen came at a community college in Queens. KeyMe placed its first prototype where regular people could approach it without a founder explaining what to do. Marsh and the team hid around a corner. One student tried it. Then another. Word moved across the campus, and soon students were lining up to make keys with the machine's novelty designs.
The line mattered less as applause than evidence. People had encountered the product cold, understood it and wanted it. Marsh says the scene established a habit: put new services near customers early, gather genuine reactions and resist building too much before the behavior is visible.
There was plenty left to learn. Hardware needed capital before it could produce much proof, while investors preferred proof before supplying capital. For stretches, Marsh says, KeyMe had only three to nine months of runway. Fundraising crowded out long-range work. Rejection became routine.
His useful response was to ask an investor who said no for five more minutes. What, specifically, had failed in the pitch? He used the answers to explain the size of the locksmith market, the reason keys were worth caring about and the role of a purpose-built machine. “You have to be hungry for feedback,” he later said, “not be defensive.” KeyMe ultimately raised more than $150 million.
“The way I've dealt with the toughest challenges is to break problems down to bite-sized, simple goals.”Greg Marsh, on his operating method
Hardware staysThe term sheet with a missing machine
One financing offer came with a condition: abandon the hardware. KeyMe needed the money, but the kiosk was not a decorative shell around an app. It was how digital geometry became a usable object in the place customers already shopped. Marsh declined. Another investor arrived soon after, solving the immediate problem and preserving the product.
Conviction did not exempt the company from logistics. When KeyMe landed its first big retail customer, machines broke and inventory ran short. Employees from the call center, engineering and the executive team boarded planes, rented cars and drove across the country to make repairs. Failure could have ended the relationship and possibly the company. The improvisation became institutional memory: scale is simply a larger number of specific things that can go wrong.
The kiosk footprint
Reported locations from contemporary interviews and KeyMe company materials. Dates reflect when each figure was published.
The result now sits in the fluorescent ecology of American errands. KeyMe kiosks appear inside retailers including Walmart, Kroger, Menards and Rite Aid. Current company materials put the network above 8,000 machines. KeyMe says it serves more than ten million customers a year through kiosks and locksmith services.
The technology also widened its vocabulary. The first machines focused on simple brass house keys. Later versions learned office keys, vehicle keys and transponders, RFID cards and fobs. Customers can choose to store a scan and return without the original. The company added a network of locksmiths for lockouts, installation and repair - the work a kiosk cannot perform, because doors are stubbornly attached to buildings.
A different seatKnowing which job to keep
Marsh eventually faced a subtler product decision: himself. A founder can become identified with the chief executive title long after the company needs a different kind of leader. Marsh has said that taking KeyMe to its next stage required experience running a bigger organization that he did not yet have. He handed the broader job to James Moorhead and became Founder & President.
The present division is clear. Moorhead leads company growth and continuous improvement. Marsh leads kiosk vision, business development and retailer relationships. He remains close to software, hardware and research, the junction where the original idea still lives.
That move reveals the temperament running through his story. Marsh likes measurable priorities. He has said that anyone in the organization should be able to name the top three and explain how they are measured. He breaks hard problems into smaller goals, writes them down and rallies people around execution. His public anecdotes repeatedly return to feedback: professors' office hours, investors' objections, customers approaching a kiosk without instruction.
He also gives credit with specificity. Early investor Ravin Gandhi met him through Columbia's Greenhouse Program and backed the proof of concept. Investor and founder Amol Sarva once spent two hours opening his contacts and making retail introductions before KeyMe had secured its first major accounts. Marsh says that generosity now informs how he responds when newer founders seek help.
The ordinary objectWhat a key can teach
In 2025, Marsh said the company expected to produce more than ten million keys that year. The number is impressive mainly because each one is singular. A key is a bespoke product cut for one lock, manufactured at retail speed by a machine expected to work for an anxious stranger. Software can scale by copying itself. KeyMe's code must command metal, cutters, blanks, motors and a supply chain.
That supply chain is now one of the constraints. Marsh has described balancing strong retailer demand against specialized assembly, tariffs and globally sourced parts. The response has included more suppliers, redesigned components and added production capacity. After years of teaching machines to read keys, the company is still learning how to make enough machines.
The founding frustration remains visible inside that complexity. A person on the wrong side of a door does not want a robotics lecture. She wants certainty about price, time and whether the next key will turn. Marsh's wager was that sophisticated machinery should make the encounter feel simpler, not more technological.
There is a charming disproportion to it all. A billion-dollar portfolio led to truck hardware; truck hardware led to a borrowed lab; a locked apartment led to thousands of yellow kiosks. Marsh did not invent the key. He noticed how much inconvenience had gathered around it, and decided the smallest object in your pocket deserved a larger idea.