Park Jun-hyuk, founder of mAy-I, poses after an interview with The Korea Times at the company's office in Gangnam District, Seoul, March 13. Courtesy of mAy-I

Internet users these days often come across cookie notices informing them that an online store uses tracking technologies to collect users' preferences and other information useful for analyzing visitor data and providing optimal environments for visitors.

While these technologies have become common among online stores, many offline stores still make decisions about store layout or operations without tangible data to support them. This is because it is difficult for store operators to track every visitor and collect data on their age, gender, preferences and other useful information.

MAy-I, an artificial intelligence (AI) startup led by founder Park Jun-hyuk, offers a solution to these challenges through its product, mAsh — an AI-based analysis system that uses closed-circuit television (CCTV) footage to extract useful visitor data for offline stores.

"One of our goals is bringing performance-based online marketing activities to offline stores," Park said in an interview with The Korea Times on March 13. "We analyze video footages from existing CCTV cameras installed in offline stores using AI and provide clients with data on who visited when foot traffic peaked and how customers moved through the space."

The company was founded in 2019 as an AI-based video analysis company. After finishing a Series A funding in October 2023, the company has so far secured 8 billion won ($5.46 million) of investments and recently launched a pre-Series B funding round with a goal of attracting an additional 5 billion won.

With its product mAsh, the company won an Innovation Award at CES 2024 and posted an over 200 percent average annual growth in revenue during the past five years. Currently, mAy-I's AI engine analyzes over 25,000 hours of CCTV footage every day.

In 2024, mAy-I analyzed 331 client stores' CCTV footage, up 32 percent from a year earlier, and the clients include a number of major businesses in Korea, such as CGV, SPAO, Hyundai Motor, Amorepacific and several big-name tech companies.

One of the model cases of using data collected through mAsh was CGV, as the product addressed the multiplex theater chain's difficulties in collecting precise viewership data useful for its advertisers.

"For example, when four people go to the movies at a CGV theater, it is usually just one person who makes the reservation," Park said. "That means CGV only has information about the person who booked the tickets — not the actual audience members. When CGV tries to sell screen ad slots to advertisers, the company needs to define who actually saw an ad or other clear insights to the actual audiences for a more efficient sales pitch to advertisers."

Park added that CGV expressed its satisfaction with mAsh, as it replaced resource-consuming and inaccurate exit surveys or other conventional viewership data collecting processes.

The interface of mAy-I's mAsh AI solution / Courtesy of mAy-I

He also noted the case of Hyundai Motor's showrooms. In recent years, carmakers' offline showrooms have transformed into a place for providing visitors with experiences related to cars and many purchase deals are signed online, thus making auto brands pay greater attention to intangible data on customers' preferences.

"These showrooms are evolving into flagship locations where brands offer curated experiences, such as in-store cafes or sections introducing the brand's history," Park said. "However, such spaces rarely generate usable data since there are often no payment records or other data points to track, thus making our solution more efficient for them."

The key technology that enables this analysis is mAy-I's AI engine for analyzing CCTV footage, which often has low resolution and limited view.

According to Park, if a human can estimate the gender and age group of a person in CCTV footage, then mAsh is also capable of analyzing it effectively. The company also offers a compact device for environments where uploading video footage online is difficult, allowing stores to run the analysis on-site without needing to connect to mAy-I's headquarters.

To address privacy and security concerns, mAy-I converts individuals captured in the footage into anonymized entities and deletes the original videos before running the AI analysis. Through this approach, the company recently obtained certifications under the European Union's General Data Protection Regulation and the AI Act, laying the groundwork for its global expansion.

Since last year, mAy-I has been providing its service to the flagship store of a world-renowned brand in Europe and occasionally offering analysis to several stores across Europe and other East Asian countries. Park said the company expects to launch its service in the United States this year, aiming to expand in regions with a strong understanding of the value of data.

"We're proud to be an AI company that goes beyond just developing technology and offers a tangible service and generates real revenue from it," Park said. "Until now, we've focused on advancing our technology and product, but going forward, we plan to devote more time to generating revenue and creating real business impact. Starting this year, we'll be scaling up our business and creating references for the global market. If we feel fully prepared, we may even begin preparing for an initial public offering."