Contactless care robots for an aging world with Norbert Health's Alexandre Winter

Alexandre Winter is the CEO and co-founder of Norbert Health, an AI-driven care automation company deploying autonomous robots with contactless, medical-grade monitoring in post-acute care facilities.

You've spent most of your career in computer vision. What's the thread that connects your early work to founding Norbert Health?

I've always been fascinated by the idea of getting machines to see and sense their environment. I started my studies in the late '90s and picked my engineering school because it had a computer vision lab. I did my master's and Ph.D. in the field, then worked in the defense industry because that was one of the first real applications of computer vision: guiding missiles, rockets, and flying objects. I loved the technology, but I was frustrated by the pace. My last project was a collaboration with an aerospace company where the delivery date was 15 years out. That's when I decided to go for something more agile.

In January 2000, I co-founded my first company. We built the first image search-by-content engine and tried to use it for e-commerce. We had a Palm Pilot with a plug-in camera and a cradle to connect to the internet. You'd take a picture of something, send the image to our cloud, and we'd recognize the object and find similar items for sale. It worked well, but there wasn't a real use case yet. We ended up finding our market in law enforcement investigations, building the first investigation engine based on image similarity and recognition. We sold it to federal agencies in D.C. and police forces around the world. The company became profitable and was eventually acquired by a Japanese firm.

After that, in 2012 I co-founded Placemeter, which used cheap, autonomous devices to analyze traffic and activity in cities at scale. We went through Techstars NYC, raised $12 million and were acquired by Netgear for the Arlo camera division in 2016. At Arlo, we were trying to figure out what to do with all these home security cameras that people put up but never really used. That's when we started thinking about healthcare. Around 2019, we realized healthcare was rapidly moving into the home for demographic and structural reasons, and it needed to be automated. So we left Arlo and started Norbert Health.

What does Norbert Health actually build?

We're an AI company. We build multimodal, multispectral AI models that can measure what's happening inside the human body without any contact, understand what's happening in the environment, and interact with people. We built a small module, a plastic box with a camera, an infrared sensor, a radar, sound capabilities, and compute power. You attach it to any standard off-the-shelf wheeled robot, connect to the robot’s battery, and that otherwise simple robot becomes a smart care assistant.

These robots carry our module around post-acute facilities like nursing homes and rehab centers, where patients stay for a month after a major operation or four to 18 months for chronic diseases. There's a doctor on site, but there's no physician on the floor all day. Medical monitoring is handled remotely by care management companies that track vital signs, activity metrics, and notes from nurses. They follow each patient's health over time and intervene when something looks wrong. They're doing preventive care for frail people, which is the biggest pain point in healthcare today. Our health system is mostly curative. The preventive side is vastly underused, and for elderly or chronically ill patients, the consequences of that gap are felt much sooner.

How do the robots work in practice on the floor of a facility?

Our robots roam the floor of nursing homes four, five, six times a day. They know who's who because they can recognize faces. So they know when a resident is mobile and walking in the hallways, having social interactions in the day room, or if someone who used to be in the day room suddenly stops leaving their room. Those are events worth reporting to care managers. We provide our customers with monthly reports and real-time alerts. If we detect an abnormally high or low heart rate, a spike in temperature, or a significant behavioral change, we send an alert right away.

Our customers don't care about seeing raw infrared or radar images. They care about medical-grade vital signs that are FDA cleared so they can use them for assessment and diagnosis. They also want contextual and behavioral information: What is the patient doing when we're measuring them? What are they doing all day? It's really about sensor fusion and a holistic understanding of the patient, not individual data from a particular sensor.

Contactless vital sign monitoring sounds extraordinarily difficult. What was the hardest technical problem to solve?

Understanding what's happening inside the human body, measuring five vital signs without contact, in an uncontrolled environment, with medical-grade accuracy. That speaks for itself. We spent the first three to four years of the company working on that exclusively. People originally said it was impossible, especially blood pressure and SpO2. But it works. We've cleared two of our vital signs with the FDA and have three more to clear. We know we can measure even more from the human body by adding different kinds of time-of-flight sensors and other frequency radars, so we'll always be pushing that forward.

Right now, everything revolving around the robot's interaction with its environment and people is also a major focus. We combine RGB cameras with radar information, for example, so there are no off-the-shelf models for that. We build those ourselves, but we also use commercial chatbots and customize them for the healthcare context. Studying how people interact with robots in a healthcare setting has been fascinating. It's completely different than a robot helping you find your way in a mall or a museum. These are more serious conversations with people in very different mindsets.

You mentioned the social dimension. How do patients and nurses actually respond to robots?

It's been really interesting. Robots in a healthcare environment can be a little scary at first, but they also provide a presence that's different from a human presence. For residents in nursing homes who are lonely, the robot becomes a social companion in addition to being useful. That role is becoming more and more important for us.

And our number one fans when we deploy are always the nurses. They're overworked. You might think they'd feel threatened, like robots are going to steal their jobs, and they'd be kicking them into closets. Not at all. They know their patients need more care and more presence, and they're genuinely happy the robots are there to help.

Healthcare can't function without more people, but there aren't enough of them. Where does Norbert go from here?

Healthcare will not work without robots. It's impossible. There's a demographic imbalance, and even if you could find enough people willing to work in healthcare, there simply aren't enough of them. You need robots across the board. The question is where do you start, how do you make them useful, and how do you find financing for them.

We found our angle with care management in post-acute facilities. There's financing for it, there's a strong need right now, and it's working today. We're covering about 500 beds, with each robot serving roughly 50 people. We're about to deploy another 150 robots over the next three to four months, which will bring us to nearly 6,000 beds. We're already in discussions with high-profile partners about pilots in the broader healthcare system, including hospitals and outpatient facilities. It's always the same problem: There aren't enough nurses, and the more they can outsource tedious, repetitive tasks to a robot, the happier they are.

What does Norbert's fundraising picture look like right now?

We're raising a $5 million round. We have a term sheet on the table and are still looking for about $1.5 million in co-investments from people who know the healthcare world and believe technology has to change it rapidly. We're looking for angel investors or small funds. Down the road, financing will become even more important. 

Once we deploy those 150 robots and reach a solid amount of ARR, we'll be in a strong position to raise a bigger round. That bigger round is where we want to have enough capital to go after the wider market. We've found the right wedge, and we're collecting data that nobody else has: real-world interactions between robots, nurses, and residents in complex environments on a daily basis. That data is extremely valuable.

SPONSORED

Infinite Frontiers is made possible by HAUS, the leading public relations agency for deep tech startups and scaleups.

Learn more