Making the physical world as programmable as a web app - any hardware, one stack.
On a given night at UBS Arena on Long Island, a fan checks the team app to find the shortest bathroom line. Somewhere in Italy, an industrial machine flags a part before it fails. On a boat, a sonar quietly learns to tell fish from noise. None of these things look related. All of them run on Viam.
Viam is a software platform for the physical world. It connects to cameras, motors, sensors, and machines, then hands engineers a clean, modular way to program them - the same way you would build and ship a web service. Robots, in Viam's telling, are just apps that happen to have arms.
The company is roughly 260 people, headquartered in New York with an onsite robotics lab, and backed by $117 million. It is not selling robots. It is selling the thing that has always been missing underneath them.
Here is the open secret of building anything that moves: the interesting part is maybe ten percent of the work. The other ninety is plumbing - drivers, data pipelines, fleet updates, the grim ceremony of making a camera talk to a motor talk to the cloud. Teams build it from scratch, every time, and then build it again for the next device.
The result is predictable. Promising hardware ideas die not because the idea was wrong, but because shipping it took eighteen months and a small army of specialists. The software world solved this problem two decades ago with platforms, package managers, and shared infrastructure. The physical world, somehow, never got the memo.
Viam's pitch is that going from first prototype to a global fleet should not require a rewrite. Usually, it requires three.
Eliot Horowitz spent over a decade as co-founder and CTO of MongoDB, turning the unglamorous problem of storing data into a company worth billions. The lesson he took away was not about databases. It was about complexity: give engineers a clean abstraction over a messy problem, and they will build things nobody predicted.
In 2020 he saw the same mess he had seen before, this time in hardware. Fragmented tools. No shared layer. Every team starting from zero. So he started Viam to build the missing stack - and, true to form, made much of it open source so engineers could bring their own hardware and skip the lock-in. He has half-jokingly called it a kind of "WordPress for robotics."
It is a contrarian bet. Plenty of money has chased flashy humanoid robots and self-driving cars. Viam went the other direction - toward the boring, durable layer underneath all of it. Boring layers, as it happens, are where the durable companies tend to hide.
Viam breaks the work of building a smart machine into reusable pieces. You describe your hardware as modules, write logic against them in familiar languages, and let the platform handle the parts that used to eat your quarter.
Connect to any sensor, camera, motor, or machine and program it with standard software practices. The robot becomes configuration, not a science project.
Deploy, monitor, and update software across thousands of distributed devices remotely - over-the-air, and resilient when the connectivity is bad.
Capture data at the edge, train and run models, and get one unified view - for predictive maintenance, safety monitoring, and visual analysis.
A largely open-source library of control and motion modules the community can publish and reuse. Bring your own hardware; skip the lock-in.
A platform's credibility lives in what people actually build on it. Viam's customer list reads like a trivia night, which is precisely the point - a stack that only fits one industry is not a stack, it is a product.
At UBS Arena, Viam ties together cameras and point-of-sale systems to watch concession and bathroom lines, then routes fans to the shorter ones through the venue app. Sbarro uses it to monitor how long pizza has been sitting on the buffet. Kongsberg Discovery is piloting Viam to put AI inside marine fishing sonar. Viking Yachts builds smarter onboard systems. Transmutex, a Swiss nuclear engineering firm, and CompScience, a workplace-safety platform, round out a deliberately eclectic roster.
Engineers built a robotic sommelier on Viam: two mechanical arms meet mid-air to pour wine into a tilted glass. The platform does not judge your taste.
Viam's mission is almost rude in its simplicity: let any engineer build, deploy, and manage AI and automation across any hardware, and cut the time from months to days. The culture follows the mission - hackathons, an active Discord, a heavy open-source footprint, and an unfashionable belief that the way to win the physical world is to invite more people into it.
If AI keeps moving out of the browser and into the world - into machines, vehicles, buildings, and ports - then something has to make all that hardware programmable, observable, and updatable without a fresh team every time. That layer will be invisible to almost everyone. It will also be load-bearing.
Viam is betting it can be that layer. Skeptics will note that platform plays are slow, that hardware is unforgiving, and that "WordPress for robotics" is a big claim from a young company. Fair. But the same was said about putting a real database under every web app, and that turned out fine.
So return to that night at UBS Arena. The fan finds the short line and never thinks about why. The machine in Italy gets fixed before it breaks. The sonar gets a little smarter. The work that used to take eighteen months and three rewrites now just happens, quietly, underneath. That is the whole idea. Viam's success looks like you never noticing it at all.