The Roboticist Who Came Second - and Built the Future Because of It

Kevin Albert came second in a high school bridge-building competition. He has not let it go. That competitive itch - the restlessness of a person who wants to build things right - runs through everything from his graduate work at MIT to his years designing quadruped military robots at Boston Dynamics, through a detour into inflatable pneumatic machines, and finally to the interior of a half-finished building in San Francisco where his robots are changing how America gets built.

For most of human history, if you wanted to finish drywall, you hired skilled tradespeople who showed up with hand tools, applied compound in layers over multiple days, sanded between coats, and left you with a surface ready to paint. The process is textured, iterative, dusty, and deeply human. It has not changed in half a century - not because nobody tried to automate it, but because nobody could figure out how to put heavy machinery inside a building under construction.

Kevin figured it out. The company he co-founded, Canvas, is now the first organization in the world to have a robotic drywall finishing operation at commercial scale. That distinction came from understanding something most robotics engineers never learn: what the inside of an unfinished building actually looks like, who works there, and what they actually need.

Kevin spent time in construction before he became the person automating it. That summer job gave him something no engineering degree could: the knowledge that the problem is not just technical. It is economic, ergonomic, and demographic. Two workers retire from the trades for every one who enters. A sector that needs to build twice as much over the next 40 years is simultaneously watching its skilled workforce age out. Kevin's answer was not software, not management consulting, and not a marketplace app. It was a robot - a real one, with wheels and sensors and a spray head - that could do the job that is running out of people to do it.

From Military Robots to Mission District Maker Space

Kevin's career reads like a greatest-hits of serious robotics. He started at the Charles Stark Draper Laboratory - one of the most storied engineering institutions in the US - as a controls systems engineer. Then Boston Dynamics came calling.

At Boston Dynamics, Kevin was an advanced robotics engineer. He contributed to BigDog and LS3 - the quadruped robots that rewrote what the world thought machines could do. BigDog was funded by DARPA and became famous for its eerie, animal-like stability on rough terrain. LS3, the militarized successor, was designed to carry gear for soldiers across terrain that would stop any wheeled vehicle. Kevin helped these robots walk. That is not an abstract achievement: it meant years of working on control systems that could respond to unexpected ground conditions, dynamic loads, and failure modes that had never existed before.

After Boston Dynamics, Kevin moved to Otherlab, a robotics incubator in San Francisco where he started a robotics research group. There he ran programs with DARPA, NASA, and commercial aerospace partners. He also met the people who would become his Canvas co-founders - Maria Telleria, who became CTO, and Henrik Bennetsen, who became CBO. They worked together for over five years before they started Canvas, which means the company was not founded by strangers betting on an idea. It was founded by colleagues who had already tested each other under pressure.

Between Otherlab and Canvas, Kevin ran Pneubotics - a company building robots out of inflatable pneumatic skins that puffed into structure. Not a detour. A laboratory for thinking about robots that could share space with humans, be lightweight, safe, and take unusual shapes. The ideas gestated in inflatable form; they came out in a 1,200-pound drywall finisher.

Canvas - Where the Robot Reinvents the Process

Canvas is not a story about robots replacing workers. Kevin has been precise about this distinction from the beginning. The robots work alongside skilled tradespeople, handling the physically punishing parts of drywall finishing - the spray application, the sanding, the dust - while human judgment remains in the loop. The machine does not follow what a person does with their hands. It does something different, something only a robot can: it applies all the joint compound in one pass, using precision positioning to eliminate the iterative layering that has defined the trade for decades.

Construction productivity has essentially flatlined for fifty years. Manufacturing improved by around 1,500% over the same period. Agriculture transformed. Even retail automated. The interior of a building under construction remained, stubbornly, a hand-tool environment. Canvas changed that equation - not with software, not with an app, but with a machine that physically works on a wall.

The Hilti partnership announced in March 2023 was a significant proof point. Hilti - one of the world's largest construction tool manufacturers - agreed to assume manufacturing responsibilities for future Canvas systems based on the proven Jaibot platform. Kevin's public comment at the time was characteristically understated: "It's extremely validating that the demand for our drywall finishing robot quickly outstretched our capacity to produce this year." Read between the lines of that sentence: Canvas had more orders than it could fill. Hilti stepped in because the demand was real.

Process First. Robot Second.

Kevin Albert's core engineering conviction is clear and worth quoting directly: "One of the core philosophies in the company is that we use the capabilities of the robot to reinvent how the process is done, and not just have a robot that does what people do."

This is not semantics. Traditional drywall finishing involves multiple coats of compound applied over multiple days, sanded between each application. The reason is human - human hands can only do so much per day, the material needs to dry, and precision is hard to maintain by hand at scale. A robot does not have these limitations. Canvas's machine applies all of the material accurately in a single pass, using the robot's positional precision to eliminate the iterative process entirely. The result is not just faster - it is architecturally different from what came before.

On the 1200CX specifically, customer feedback shaped the product's most distinctive feature: compactness. Earlier Canvas systems could finish large commercial spaces, but struggled in tight corridors, smaller rooms, and multi-family residential buildings. The 1200CX, at 30 inches by 34.5 inches, can get into those corners. As Kevin put it: "Being able to be compact and get into small spaces allows them to use the robotic system for all of their walls as opposed to just some of them." That shift - from "some of the walls" to "all of the walls" - changes the economics of the whole product.

Canvas also designed out a common friction point for construction robots: site planning. You do not need architectural drawings to run a Canvas machine on your job site. The robot automatically detects the wall geometry. In a sector where paper plans and digital models often diverge from what is actually built, that matters more than any spec sheet suggests.

Built in Engineering Rooms

Links & References