Robots at Your Doorstep
Our Guest- Marko Bjelonic
Marko Bjelonic, born in 1990 in the former Yugoslavia, moved to Germany during the 90s' civil war. He pursued his education in Germany, Norway, and Australia, aiming to complete a PhD at ETH Zurich, a leading robotics university.
Inspired by notable professors at ETH Zurich, he joined a lab focused on legged robots as one of the first PhD students under Prof. Dr. Marco Hutter. During his PhD, Marko contributed significantly to robotics.
He pioneered a patented technology combining wheels and legs for quadrupeds and contributed to the first use of an artificial neural network on a legged robot, advancing Physical AI.
He was also part of the winning team in the third DARPA Robotics Challenge, which The Washington Post called the "Super Bowl of Robotics." Marko co-founded RIVR, originally Swiss-Mile, a Swiss robotics company with backing from Jeff Bezos and HSG. RIVR is at the forefront of using General Physical AI to transform last-mile delivery.
“You don’t scale robots by forcing society to adapt. You scale them by building machines people trust - machines that fit naturally into human life.”
-Marko Bjelonic
When Marko Bjelonic talks about delivery robots, he doesn’t start with technology. He starts with a childhood spent helping his father rebuild a life from scratch. Long before he was a robotics PhD or a Silicon Valley–backed founder, he was a 14-year-old refugee in Germany, writing invoices on a family computer and learning what real customer service looks like on dusty construction sites.
His father had just lost his job. Instead of giving up, the family opened a tiny construction business and Marko watched close-up how trust, hard work, and saying “yes, it’s possible” kept customers coming back. Those early lessons shaped him far more than any research paper. They taught him that technology means nothing unless it solves a human problem.
That mindset, not robotics, is why RIVR exists.
From Refugee Kid to Robotics Researcher
Marko didn’t grow up surrounded by engineers or labs. He grew up watching his parents work seven days a week after arriving in Germany as refugees from Latvia. “I saw this hardworking culture, but it was always for us, the children,” he says.
When his father lost his job, the family didn’t wait for luck. They created it. “Sometimes you just have to jump into the cold water,” Marko says. That philosophy followed him all the way to ETH Zurich, where he studied legged robots, creatures built to handle the unpredictable world humans move through every day.
But even during his PhD, he noticed something strange: “Everyone was trying to make legs stronger and faster. But humans already invented something brilliant, the wheel.” So he combined the two. His hybrid leg-wheel robot could walk, roll, climb, and accelerate. A strange design but perfect for real streets.
The Problem That Sparked RIVR
While most roboticists were dreaming of humanoids doing flips, Marko was watching food delivery dRIVRs collapse under impossible workloads. “The first mile and middle mile are predictable,” he explains. “But the last mile? That’s where the messiness starts.”
DRIVRs get tired. Packages get mis-sorted. Traffic gets chaotic. Customers get frustrated.
That’s when his engineering instincts and his customer-first upbringing collided. “You can’t achieve the experience consumers want, predictability, convenience without removing the stress from humans,” he says.
RIVR was born from that single belief: machines should take the pressure so humans don’t have to.
Why He Didn’t Build a Humanoid (Yet)
If you scroll through social media, it looks like the future belongs to humanoids that can sprint, jump, and fold laundry. Marko disagrees. “For our use case, a humanoid would be useless,” he says flatly. “It can’t cover big distances fast. It’s too expensive. Too complicated.”
His robot needed to climb curbs, navigate steps, and cruise for kilometers at speed. The dog-like form wasn’t stylistic, it was practical.
“We want to build general physical intelligence,” he says. “But to get there, we need real-world data. The fastest way to collect that data is to put thousands of robots into the world, not wait years building humanoids.”
A humanoid may come later. But not until customers actually need one.
Autonomous Vehicles Aren’t Enough
People often ask Marko why RIVR doesn’t just rely on self-driving cars. His answer is simple: “Most Americans live in single-family homes. Packages go to porches. An autonomous car can’t walk up your steps.”
The real solution is a robot + autonomous vehicle working together.
Cars handle long distances. RIVR handles the last hundred yards. That’s the synergy he’s betting on.
General Physical Intelligence Starts With Real Streets
Marko talks about data the way others talk about oxygen. Without it, robots don’t learn. With it, they become capable of navigating the unpredictable world humans take for granted.
“YouTube is great for training video models,” he says. “But for robots, we need the camera image connected to the robot action. You can’t get that from the internet.”
Once tens of thousands of RIVRs are deployed, the data they collect could surpass what YouTube uploads in a single day. That’s the scale required to build physical intelligence, not in theory, but in everyday environments.
“It’s not black and white,” he adds. “Robots get smarter along a spectrum. More data, more intelligence.”
What Happens to Delivery Workers?
Marko is blunt: “The stressful part gets handled by the robot. Humans become the support system.”
RIVR has already hired former delivery couriers as robot operators, technicians, and supervisors. “It’s less stressful, better paid, and more creative,” he says. Despite fears of job loss, he sees the opposite: a shift toward higher-skilled roles, just like in past industrial revolutions.
The Wild Story of How Jeff Bezos Got Involved
Marko’s first trip to San Francisco is now a startup legend. He knew no one. He sat in a café. He started talking to the stranger next to him. Three weeks later, an email arrived: Jeff Bezos wanted a meeting.
“If you don’t create the probability of luck, you won’t experience luck,” he says. “You have to show up.”
Why RIVR Is Built in Switzerland but Not Limited to It
ETH talent. Lower salaries than Silicon Valley. Engineers who stay for years, not months. These are the advantages of Switzerland. But Marko never wanted to stay confined there.
“From 15 people, we already had offices in Shanghai and Austin. You can’t build robotics from one place,” he says.
Today, RIVR’s footprint spans Zurich, Austin, Leeds, Amsterdam, Luxembourg, Pittsburgh, and Palo Alto.
The Long-Term Vision: A Helpful Robot In Every City
Delivery is only the first chapter.
“Once you have tens of thousands of robots in cities, they can help with anything,” Marko says. “Carry your child’s school bag. Walk with someone who needs support. Assist older citizens. Become helpers for humans.”
It’s not a sci-fi fantasy. It’s a logical evolution once robots are everywhere.
