Zenus Founder Spotlight

How Panos Moutafis Is Overcoming the Challenges of Computer Vision to Deliver Ethical AI

When Panos Moutafis immigrated to the United States in 2011, his plan was to become a professor. But while earning his Ph.D. in computer science from the University of Houston, fate intervened. The University successfully filed for a patent based on the work he was doing in computer vision, facial recognition, and artificial intelligence, paving the way for him to start his own business to commercialize the technology he had developed.

In 2015, Panos founded Zenus, a company that provides ethical facial analysis to help organizations better understand human behavior so that they can create better experiences for their customers. At a time when attendance at live events is returning to pre-pandemic levels and people are craving in-person interactions, Zenus’s smart camera and AI-powered software overcome many of the challenges associated with computer vision. In fact, they’re being used everywhere from airports and amusement parks to retail stores and trade shows.

“In-person interactions matter,” says Panos. “They’re critical for building trust, fueling innovation, and giving people the opportunity to learn and grow. We develop ethical facial analysis  and artificial intelligence to help our customers better understand those interactions so that they can personalize and optimize them and ultimately drive better results.”

Overcoming the Challenges of Computer Vision

When Panos first launched Zenus with co-founder Rakshak Talwar, the company was focused on building a cloud-based facial recognition platform. Customers would send Zenus video from their events in real time to identify attendees and make the check-in process faster and easier.  

“What we realized is that most of the companies we were working with didn’t have any expertise when it came to integrating computer vision with video applications and other systems,” says Panos. “Plus, sending video to the cloud for analysis is difficult and expensive because of the amount of bandwidth it requires — we’re talking terabytes of data every day. And then there were all of the potential privacy and security issues that come with transmitting all of that video data to the cloud.”

In 2020, Zenus pivoted its approach by launching an on-premises edge processor with six CPU cores and 384 GPU cores embedded in it, giving it a huge amount of computing power. Customers can plug cameras into the device to analyze video locally, capturing an array of useful data points in the process. As a result, all the device needs to transfer to the cloud is the aggregate metadata from that analysis — typically a single megabyte of data per hour. In doing so, Zenus eliminates both the bandwidth and security issues it previously faced since the data isn’t tied to any individual. And, since it can be set up within a matter of minutes, end users don’t need any expertise to use it.

Plus, once the metadata is transferred to the cloud, Zenus’s customizable, AI-powered software converts it into useful visualizations and other outputs to help customers solve any array of business problems.

Putting Zenus to the Test

One place Zenus’s technology is being deployed is at airports, where officials wanted to understand if travelers were actually still using flight information displays or just relying on their phones instead. To answer that question, Zenus integrated its devices into two flight information displays at a major airport to measure how many people look at them in an average month.

“We quickly determined that tens of thousands of people were looking at the displays, meaning that they were still valuable,” says Panos. “But then we took our analysis further by layering in all of the data that our technology captures, including demographic information, sentiment, and dwell time, among many other data points. That data led to some important inferences about travelers’ experience with those information displays.”

What Zenus noticed was that older travelers generally spent more time looking at the flight information displays and were generally less happy doing so. The underlying assumption was that they likely weren't able to see the displays as easily and may not have been able to find the information they needed as quickly as others. Zenus decided to test the hypothesis by working with the airport to run an experiment. They upgraded one of the informational displays so that it was easier for people to see and understand while leaving the other unchanged. They then measured how people were interacting with the displays for another month.

“The improved display had much better sentiment scores and lower dwell times with people of all ages, especially older travelers,” says Panos. “After sharing the findings, the airport updated all of its displays to the new format, creating a better traveler experience. It’s obviously a very small thing on its own, but when you think about all of the other applications for this kind of technology, you can begin to imagine how the compounded effect this could have on so many of our day-to-day experiences.”

Another common use case for Zenus is the $65 billion event sponsorship industry, where exhibitors regularly spend tens of thousands of dollars on booths in the hopes of attracting potential customers. Of course, how they design those booths has a big impact in terms of how many conversations individual exhibitors will have with prospective customers, how many demos they’ll run, and how well the information they share will be received.

Zenus helps companies measure and understand what booth designs, types of demos, and other factors lead to the best outcomes at trade shows and other events. As in the airport example, it does this by measuring everything from impressions and dwell times, to demographics and the overall sentiment of booth visitors, such as whether they look happy or surprised, or even if they laugh while interacting with the booth.  

To help exhibitors understand the effectiveness of their booths, Zenus introduced a metric called stop rate, which refers to the percentage of people who not only stop at a booth but also linger. The higher the stop rate, the more leads an exhibitor can expect to capture. Armed with this data, Zenus can help exhibitors identify which features their booths and demonstrations should incorporate to maximize stop rates and ultimately win more business.

“We help event exhibitors measure their ROI and optimize their efforts,” says Panos. “By being able to quantify the impact their spend is having, they can make better decisions about how to allocate their budget dollars.”

An Ethical Eye on the Future

Today, Zenus is a team of 10 people and growing. “We wouldn’t be where we are without the incredible hard work and dedication of everyone who works here,” says Panos. Headquartered in Austin, Panos expects 2023 to be a year of massive growth. The company raised its Seed round in late 2022 and is already planning for its Series A round later this year.

“We love working with the Moneta team,” says Panos. “We could have partnered with a lot of other investors, but Moneta stood out for a number of reasons. They’re former operators themselves, so they have a real understanding of businesses and are able to help us solve real problems. They never hesitate to roll up their sleeves and get their hands dirty. Plus, everytime we talk to them, there’s always a lightbulb moment where we get some kind of new insight or idea that helps us work through whatever challenge we might be facing.”

Looking ahead, Panos says Zenus will expand into retail. The company is also focused on building sales and marketing pods for each market it’s going after so that it’s able to speak the same language as its customers in each individual segment. Panos says that the company will also focus on building partnerships with existing players in the industries it’s targeting to help accelerate penetration. He’s also looking forward to utilizing their device’s capability to be updated remotely, allowing more features and functions to be added over time.

“We’re not just building a facial analysis solution,” says Panos. “Eventually, it’s going to become a scalable platform for edge AI. And, perhaps most important of all, it’s always going to be ethical because it will always protect people’s right to privacy.”

Read these next