[Korean Startup Interview]MAZE: “Developing Vision AI to Remember Human Personas, Aiming to Lead Global Offline CRM”


  • Funding Stage : Bridge Round
  • Raising amount : KRW 300 M
  • Desired Fundraising Timeframe : 2Q 2025
YouTube video

What is the most critical asset for companies marketing to customers today? Undoubtedly, it’s “data.” Information on which customers showed interest in what products, what they searched for, and how long they lingered on a product page constitutes a company’s most valuable resource.

The easiest way to gather such data is likely through “online” channels. In an era where everyone owns a smartphone, the digital footprints people leave online become data in themselves. However, the online data market is dominated by global Big Tech firms operating SNS platforms, while the e-commerce space is crowded with giant retailers, making it tough for newcomers. Including financial firms’ payment data, the online data market is essentially a monopoly stage for large and global corporations.

But what about the offline market? The data from physical spaces—where people meet friends, shop, eat, browse, and make purchases—is just as vast and valuable as online data, yet no dominant player has emerged. A startup is extracting meaningful insights by analyzing customer behaviors, reactions, and preferences in this offline market. That company is MAZE, the focus of today’s article.

Take pop-up stores as an example. These events allow companies to gauge customer interest and reactions to new products. Until now, the quantitative data from pop-up stores has been limited to metrics like visitor counts or gender ratios, often collected manually with clickers throughout the event. If data on which age groups lingered at specific products, how long they stayed, and their facial expressions or reactions could be quantified, it would become essential for marketing. MAZE has developed a solution called “Live Review” to make this possible. Live Review is not only for pop-up stores but also for exhibitions, trade shows, and any business or individual operating offline stores.

A counterargument might arise: couldn’t such analysis be done later using CCTV footage? Indeed, some companies analyze customer reactions from CCTV recordings. However, CCTV requires capturing, transmitting, storing, and processing massive data volumes, incurring significant costs. Privacy concerns necessitate de-identification during analysis, adding complexity. Post-analysis, ensuring no data leaks requires deleting transmitted footage, demanding further attention and expense.

MAZE addresses this by capturing key customer features—facial structure, gender, age, height, etc.—along with their movements and reactions at specific products or booths, storing them as “numbers” without saving video. Instead of storing and analyzing large video files over days, MAZE digitizes behaviors instantly, enabling immediate data analysis. According to MAZE’s founder and CEO, Ki-sun Song, “We don’t store customers as video; we generate real-time data the moment they act.”

If this technology is feasible, companies have little need for costly hardware, expensive analytics equipment, or pricey software. One of MAZE’s key strengths is offering “Live Review” at a reasonable price. Song explains, “Storing and training AI on countless videos requires vast datasets, storage, and high-salaried engineers. Theoretically, costs could spiral endlessly, but such high-cost models are unsustainable for businesses.”

Despite its affordable pricing and technical strengths, questions remain. Offline spaces see countless people with similar facial and physical traits. How can MAZE assign distinct numerical values to differentiate personas? The answer lies in CEO Ki-sun Song’s background.

Song graduated from Seoul National University’s Department of Mechanical and Aerospace Engineering and earned a Ph.D. in aerospace engineering from Georgia Tech. He interned at Toyota and Siemens before working as a lead researcher at Hyundai Motor Company, where he led efforts to quantify aerodynamics in vehicle driving. His expertise in turning complex information into numerical data is evident. Song says, “We’re not an AI company. Human features and behaviors can be mathematically modeled from the start, and Live Review quantifies them the moment a person appears.”

MAZE’s technology isn’t just for businesses. Public institutions like police, tracking suspects or finding missing children, can use “Live Review” efficiently. It eliminates the need for officers to spend days tracing suspects’ movements across multiple videos. Quantified suspect features enable instant tracking. The same applies to locating missing children. Extending this to identify object features could open new markets.

With this technology, MAZE, in its second year since founding in 2024, achieved over KRW 1.1 billion in revenue and aims for KRW 5 billion this year. After securing seed funding last year, MAZE plans a seed bridge round mid-this year to fuel aggressive sales and R&D. WOW Partners, recognizing MAZE’s agility and grounded business vision, selected it as one of WOW NEXT’s 1st Cohort.

Investors curious about the vision of CEO Ki-sun Song, who left stable roles at global and Korean giants to start MAZE before age 40, should explore MAZE’s plans in this year’s funding round.

What problem is the MAZE team trying to solve?

The online marketing market has grown rapidly since smartphones emerged. In the era of one phone per person, online platforms track individuals via codes, building digital personas for precise marketing, converging into online sales.

Naturally, businesses want similar marketing and sales capabilities offline. Many Big Tech firms analyze CCTV footage to provide insights, but despite numerous attempts, the high costs yield limited value, making it challenging to advance offline sales based on customer personas.

MAZE tackles this by developing core technology to build offline customer personas, offering solutions for real-world sales akin to online personalized tracking.

How does MAZE solve this problem?

We reimagined technology with a new paradigm. By discarding traditional video analysis approaches and adopting a fresh perspective, we drastically reduced analysis costs, developing cutting-edge technology usable at a reasonable price.

Given the real-world need to recognize and remember new people instantly, we developed algorithms to handle this and an architecture to integrate them, enabling our service to identify customers across different times and places.

What are our competitive advantages and technical strengths compared to competitors?

First, our reasonable pricing. MAZE’s solution works with standard CCTV, requiring no specialized equipment or large-scale hardware since it doesn’t store videos.

Second, ease of adoption. Once installed, MAZE’s solution converts customer data into proprietary numerical formats without image storage, requiring no additional client effort.

Third, unique value proposition. Data companies must answer whether their service boosts revenue or cuts costs. MAZE provides tangible answers to these questions.

Finally, MAZE can theoretically be deployed anywhere without relying on video data, a significant strength.

What products/services does MAZE offer, and what is their current status?

We offer a service called “Live Review.” Currently, it’s provided for offline experiential events like pop-up stores, exhibitions, and trade shows, where many visitors are expected in a short time, requiring significant resources.

A unified dashboard is provided at the event’s start, offering real-time data on visitor counts, repeat visitors, gender ratios, and dwell times during the event.

After the event, by the next business day, inputting event costs yields metrics like CPA, CPM, and CRC. We also compare spending against the average of all clients using MAZE’s service for objective analysis.

We provide PDF reports, raw data downloads, and other services.

What is the target market size, and who are the core customers?

The big data market is valued at USD 17.86 billion (KRW 25 trillion), and MAZE aims to initially target USD 15 million (KRW 21 billion) in revenue.

All companies needing metrics from offline events like trade shows, exhibitions, and pop-up stores to drive sales are our initial customers.

Assuming 20 trade shows annually, 7,000 participating companies, a service fee of USD 357 (KRW 500,000) per event, and a 30% conversion rate, we project USD 15 million (KRW 21 billion) in annual revenue.

What is MAZE’s business model?

We’ve found product-market fit in the offline advertising and customer data market, previously deemed unquantifiable, charging around USD 357 (KRW 500,000) per booth for experiential events.

We ask for understanding as our new business model, set for launch this year, cannot yet be disclosed.

What are the MAZE team’s achievements?

We raised USD 357,143 (KRW 500 million) in seed funding, were selected for TIPS, and achieved first place in the 2024 Early Startup Package (Suwon University) and top graduation honors (Startup Promotion Agency Award) from KISED.

In 2024, we filed 20 patents, registered three, participated in IBK Changgong Daejeon 6th Demo Day, D.CAMP’s March 2024 D-day, and earned venture enterprise status, Gyeonggi Province Promising Small and Medium-sized Enterprises, ISO9001:2015 certification, and numerous awards.

In 2024, less than three years since founding, we recorded over USD 7.86 million (KRW 11 billion) in revenue.

What is the MAZE team’s competitiveness?

Led by a CEO with bachelor’s and master’s degrees from Seoul National University’s Mechanical and Aerospace Engineering and a Ph.D. from Georgia Tech, alongside five top-tier CxOs, our entire staff sacrifices for the company. United by passion, grit, and collective intelligence, our board and employees create a new world of customer personas with unmatched effort and results.

Why should we receive investment? Three reasons:

First, we’re revolutionizing the offline marketing market with technology, achieving early commercialization.

Second, our CEO and team exhibit top-tier capabilities and teamwork, working as “performers” for customers with the support of many.

Third, while Big Tech dominates online data with SNS platforms, MAZE pioneers the offline market with innovative technology, poised for partnerships with leading online firms, offering limitless potential.

Anything else you’d like to convey?

We sincerely thank you for this opportunity to introduce MAZE. We hope customers and investors will show great interest in our journey.

Share

Leave a Reply

Your email address will not be published. Required fields are marked *