He built a database the internet runs on. Now he is writing the software layer for motors, arms, sensors and the physical world - and betting that robots are finally ready.
A chess-playing robot sat on a table in a New York apartment during the worst stretch of the pandemic. Next to it, blueprints for a cat feeder that fed on schedule and a sprinkler smart enough to skip the rain. The person assembling them had just walked away from a company worth tens of billions of dollars. Most people in that position buy a boat. Eliot Horowitz built one for the boat instead.
Today he is the founder and CEO of Viam, a New York company building the software platform for robotics and physical automation. The pitch is deceptively plain: make programming a motor feel like programming a website. Behind it sits a conviction Horowitz formed while soldering hobby projects at his kitchen table - that the machines were ready, and the code was not.
"The hardware people can run laps around the software."
That single sentence is the whole thesis. Cameras got cheap. Chips got fast. Motors got precise. And yet, in Horowitz's reading, building a working robot in 2020 still felt like building a website in 1999 - bespoke, brittle, and reinvented from scratch every single time. He spent a year trying to find the platform that would fix it. When he could not find one, he did what he has done his entire life. He wrote it.
The origin story is almost suspicious in its tidiness. Horowitz, born in New York in 1981, started coding at four. By nine he was writing software for his mother's medical practice - real software, doing real work, for an actual office that depended on it. There is a particular kind of engineer who treats a computer less like a tool and more like a native language. Horowitz is that kind.
He went to Brown University for computer science, then landed at DoubleClick in its research group. That detail matters more than it looks. At DoubleClick he met Dwight Merriman, the company's co-founder and CTO. Years later, that relationship would seed the most important chapter of his career. In the meantime, he co-founded an internet shopping search engine called ShopWiki in 2005, building the crawling and data-extraction algorithms himself and serving as its chief technologist. The work earned him a spot on BusinessWeek's Top 25 Entrepreneurs Under Age 25 in 2006. He was, by then, an experienced founder and not yet old enough to rent a car without a fee.
In 2007, Horowitz reunited with Merriman and Kevin P. Ryan to start a company called 10gen. The plan was ambitious and slightly heretical: build a database that fit the way modern web developers actually thought, rather than forcing them back into rigid rows and columns. Horowitz wrote the core codebase. The product was called MongoDB.
What followed is one of the defining arcs of modern infrastructure software. MongoDB's first public release landed in 2009. In 2013 the company took the product's name as its own. In 2017 it went public on NASDAQ, raising roughly $192 million in an IPO whose shares jumped more than thirty percent on day one. The market capitalization would go on to climb into the tens of billions. For thirteen years, Horowitz led product and engineering as chief technology officer - long enough to watch "NoSQL" go from a punchline to a default.
"Everything in software has fundamentally changed in 25 years. Robotics tooling had not."
He stepped down as CTO in July 2020, staying on as a technical advisor. Most founders would have called that a complete career. Horowitz treated it as a warm-up.
The Viam idea did not arrive in a boardroom. It arrived through hobby projects built, in part, with his own family during lockdown. An automated cat feeder. A chess-playing robot. A backyard sprinkler that knew better than to water in a storm. Each one taught the same lesson from a different angle: the parts were available and affordable, but stitching them into something that actually worked was absurdly hard. There was no MongoDB for machines.
So Horowitz reverse-engineered the problem the way he reverse-engineers everything - through abstractions. Viam defines a clean API for each piece of a robot. A motor is a motor. An arm is an arm. A camera is a camera. Each gets a gRPC interface that behaves the same whether it is bolted to a factory line or a boat hull. On top of that sits the cloud: data capture, machine-learning pipelines, fleet management, over-the-air updates, and the kind of remote diagnostics that web developers have taken for granted for two decades and roboticists have mostly gone without.
The result is a company that quietly slipped past the word "robot." Viam now describes itself as the engineering platform for the real world, and its customers prove the range. At the New York Islanders' UBS Arena, Viam ties cameras to point-of-sale systems so a fan's phone can point them toward the shortest concession line. A boating outfit called Canyon Runner worked with Viam to build a tool for an Atlantic expedition, tracking location, speed and water temperature in places with almost no connectivity. Predictive maintenance, robotic sanding, marine sonar, restaurant automation - the through-line is not a kind of machine. It is the software underneath all of them.
In March 2025, Viam raised a $30 million Series C led by Union Square Ventures, with Battery Ventures and the Italian group Neurone joining. It followed a $45 million Series B the year before and pushed total funding to $117 million. Albert Wenger of Union Square Ventures framed the bet plainly, calling Viam, under Horowitz's leadership, "that bridge between the physical world and AI."
Horowitz's own framing is less about machines and more about people. He keeps returning to the idea that far too many would-be builders assume a robotics business is impossible, and that the assumption itself is the bottleneck. Lower the barrier, he argues, and you get the same explosion of creativity that cheap web tooling unleashed a generation ago. His description of success is almost modest: "Lots of mundane problems solved and lots of really interesting novel problems solved and humans doing more interesting things." It is the engineer's version of utopia - less drudgery, more invention, and a very clean API in between.
There is a pattern here worth naming. Twice now, Horowitz has looked at a category everyone considered mature - databases, then robotics - and decided the real problem was that the software underneath was a generation behind the demand on top of it. He was right the first time. The $117 million on the table is a wager that he is right again.
The hardware people can run laps around the software.
Software tooling in the robotics space was not good enough. In the last 25 years, everything else has fundamentally changed.
There's a gRPC API for every component you need - motors, arms, the pieces of robots that people actually use.
So many people think starting a robotics business isn't possible. We need to make it clear that's not true.
At the Islanders' UBS Arena, Viam links cameras and point-of-sale data so a fan's phone steers them to the least crowded concession stand.
Boating company Canyon Runner worked with Viam on an expedition tool tracking location, speed and water temperature where connectivity barely exists.
Predictive maintenance, robotic sanding, marine sonar, restaurant automation - all running on the same component APIs and cloud backbone.
He wrote the very first line of MongoDB's core code - the database now sits under a sizable slice of the modern web.
He started programming at age four and shipped software for a working medical office by age nine.
His follow-up to a multi-billion-dollar database was, literally, a chess-playing robot.
Both of his defining companies, MongoDB and Viam, are headquartered in New York City.
The MongoDB founding team was a DoubleClick reunion - he met Dwight Merriman there long before they started a company.
Viam quietly outgrew the word "robot," rebranding around the broader idea of an engineering platform for the real world.