Radius AI brings edge AI to retail, wins Innovation Award

At VentureBeat’s fourth annual AI Innovation Awards given out in July at Transform, Radius AI, based in Tempe, Arizona, won the Edge AI award for its artificial intelligence (AI) computer vision technology that equips retailers with real-time insights on queue analytics, customer counts, store layout, parking lot analytics, customer journey and employee metrics.

According to nominating committee member Andrea Huels, head of AI, North America at Lenovo, Radius AI’s computer vision solution helps retailers, particularly convenience stores, take advantage of missed opportunities, such as knowing if someone left the store without purchasing. or if the store’s product placement is ideal. The goal is to provide insights to help retailers create more sales and satisfied customers.

“The focus of their human-centric AI is to nudge staff in real-time so they can make improvements,” said Huels.

Radius AI is currently doing a 1,000-location deployment for an enterprise client, which Radius claims to be the largest computer vision project in the convenience store sector.

Collaborating on edge AI

Back in 2017, Radius AI founders Abhinav Chowdary and Jeff Cox became friends while working together at Wells Fargo in Arizona, but the two had long spoken about collaborating on a startup, particularly in computer vision.

“At the time, retailers were not thinking about the edge,” Radius AI cofounder Susan Sly told VentureBeat. “Many were contemplating how to use AI for online engagement, but not so much in terms of how it could benefit the customer experience in brick-and-mortar stores.” But Chowdary and Cox realized that convenience stores, with their already-existing on-premises security cameras, were a highly-successful segment that was underserved by AI-driven computer vision technology.

Current Radius AI CEO and cofounder, Aykut Dengi, was an engineering professor at Arizona State University when he met Chowdary and Cox in 2018. Excited by the Radius AI technology, he took a sabbatical from teaching and offered his services for free. Finally, Sly was brought in as a fourth cofounder – after two decades building large sales teams as an entrepreneur, she was serving on a Phoenix advisory board for startups and looking for her next challenge.

“Candidly, less than 2% of tech companies in the U.S. have at least one female founder,” she said. “So the opportunity to partner with a team of people who value inclusion and results was an easy ‘yes.’” In addition, Sly said she could easily see how the Radius AI technology could radically shift the brick-and-mortar landscape.

“Although I hadn’t written a line of code since 1992, I was eager to jump in,” she said. “Even in the 1990s, when I was working on an early facial recognition project, I could see where this technology would eventually take us.”

Convenience stores were an untapped opportunity

Over 29% of Americans go to a convenience store every day, Sly claimed, but as a result of the pandemic and other macroeconomic factors, there has been a shortfall of retail workers – which has led to challenges in meeting customer service expectations. That has meant an untapped opportunity for growth, she explained.

“One of the things that we care about in the store is how we can use AI — and specifically computer vision — to increase parking lot-to-store conversion,” Sly said. “Because once we can get people in the store, then they’re more likely to buy something.”

Each convenience store has its own Radius AI edge device, which, combined with the existing security cameras, allows for fast and secure data visualization. The company’s intelligence solution offers insights from the time customers enter a store, including customer counts, paying vs. non-paying customers, heat maps and a full customer journey.

Sly noted that a key differentiator for Radius is its relationships with partners that do hardware integration. That means the company can tackle the installation, the edge device and the GPUs, as well as services.

Radius AI has built out its own custom models. While there are lots of different labeled training datasets in computer vision for things like image recognition, there isn’t readily-available training data to recognize human intent.

“We have developed fundamental models that capture human motion and we interpret those actions according to the context,” Dengi said. In retail, he noted, that context is critical. For example, when COVID-19 restrictions were in place, many retailers implemented social distancing rules. While that made it more difficult for some computer vision models to properly identify things such as how busy a store was, Radius AI’s edge system was able to adapt and maintain accuracy.

Radius AI’s computer vision future

Sly said that Radius AI is currently focused on three things: Trust, ease and “wow.”

“Our goal is to be the most trusted computer vision company in the world,” she said. “We understand that it is a bold statement, but we want the adoption of our product to be easy and we work with strategic partners to figure out ways to deploy computer vision efficiently – our record for getting a location up and running is one hour and we want to beat that.”

Lastly, she said, the company wants to “wow” end users. “Our vanguard opinion is that we believe that humans and AI are better together,” she said. “Five years from now, we will be releasing products that help humans be superheroes.”