Google rarely invests in Japanese startups. Meet the exception.

Abeja is an artificial intelligence consulting company that’s making waves in the Japanese tech industry.

The startup has raised a total of US$45.3 million in disclosed funding to date, with the latest investment from coming from Google via a series C round in December 2018.

Google’s move made headlines as it was one of the company’s rare investments in Japan. The US tech titan and Abeja plan to collaborate on developing AI and machine learning solutions for retail, manufacturing, and other sectors.

Google is not the only well-known backer of Abeja. The series C fundraise included chipmaker Nvidia, air-conditioning brand Daikin Industries, and optical equipment manufacturer Topcon Corporation.

Yousuke Okada, founder of Abeja / Photo credit: Abeja

n n

Several of these investors are Abeja’s customers. Daikin has worked with the startup on using deep learning to improve its operations. For example, it could be used to help identify faults when fixing faulty air-conditioning units belonging to its customers.

Behind the rise of this fast-growing company is the story of Yousuke Okada, its 31-year-old founder.

nEngineering-savvy foundern

Okada started programming at 10 years old and majored in computer graphics in high school.

He continued pursuing his passion in college, majoring in fields that use computing power, such as fluid simulation. While there, Okada started a photo-sharing service similar to Instagram. When that business failed, he joined RichMedia, a Silicon Valley internet services company. That experience exposed him to various aspects of a business: sales, marketing, design, finance, human resources, and more.

The idea to launch Abeja came when Okada first learned about deep learning, an emerging technology that’s making AI practical in many fields.

In September 2012, he started Abeja with the help of angel investor Keisuke Tomimatsu, the former executive director of Morgan Stanley Japan Securities and currently an independent director of Musashi Seimitsu Industry, an automobile parts company.

Abeja’s unique approach of consulting for companies across different industries stems from Okada’s background in the arts.

He cited a quote from Steve Jobs: “It is in Apple’s DNA that technology alone is not enough – it’s technology married with liberal arts, married with the humanities.”

At first glance, Abeja’s projects don’t seem to be connected to each other. This year, it announced that it will work with Singapore’s public transport operator SMRT Corporation to use deep-learning technology to automate fault identification in railways.

Abeja is also collaborating with auto-parts maker Denso International Asia to analyze worker movements to improve work efficiency in its Thailand-based factories.

Set up in 2012, the Japanese startup has grown rapidly in seven years. It has helped companies to implement over 200 projects across a wide range of industries, including transportation, healthcare, logistics, and retail.

nBeyond consultingn

Abeja isn’t just throwing people at a client’s problems – it’s developing its own technology as well.

The company’s platform provides its customers with a basic AI infrastructure, which helps them cut down on efforts to create their own AI models. It was designed with data scientists, infrastructure and application engineers, and product owners in mind.

“When a company wants to adopt AI in its operations, it has to build secure and scalable infrastructure to accumulate data, train, deploy, and manage the machine learning models. It takes a long time to do this from scratch,” says Okada.

Customers can simply log in and start using Abeja’s platform quickly because it has existing deep-learning models that are trained on other datasets. A company can take a template model and customize it with data gathered from its own sources. The platform also has a proprietary way of helping customers analyze unstructured data.

More than 150 companies are now benefiting from Abeja’s platform. In factories, for instance, it could be used to analyze the work process of skilled craftsmen. The data could then serve as the basis for instruction manuals.

While Abeja serves a diverse client base, the majority of its clients are in the retail sector. The startup is assisting them in gaining insights on consumer behavior, store management, and product demand. It can even recommend pricing changes based on supply circumstances and market demand. Retailers could then use this information to make business decisions, like changing a store’s product lineups and displays.

Abeja has installed cameras in the Parco-ya shopping center in Tokyo to gather data about shoppers. It analyzes images to tease out a visitor’s age, sex, and shopping behavior, such as how often they make a trip to a store. According to a 2018 article in The Asahi Shimbun, the mall operator found that 80% of its customers were women, and most of its visitors were in their 30s to 50s.

Meanwhile, ICI Ishii-Sports was able to collect data that dispelled its assumption that most of its clients were in their 40s to 50s. It had approximately the same number of shoppers in their 20s and 30s.

nA customized strategy for different marketsn

Abeja’s target markets are Japan, Singapore and Thailand, and the US. It is preparing to incorporate its Thailand branch as well.

The company has a different strategy for each market. “For instance, in Singapore, we would like to collaborate with companies in smart city initiatives and logistics because Singapore is our showcase in Southeast Asia. In Thailand, we focus on industries such as manufacturing, infrastructure, and finance.”

Japan, on the other hand, has its own set of problems, such as low birth rate, an ageing population, increasing social security expenditure, and economic stagnation. “We believe that we can solve these problems by using technology to increase the country’s gross domestic product,” says Okada.

While Okada declined to provide Abeja’s financial figures as it’s not listed, he says that the company is seeing “double-digit growth per year.”

One of Abeja’s challenges, however, is in making its product easy to use and understand for customers with no skill sets in machine learning, deep learning, and AI. Okada says the startup is developing educational materials for companies, especially for corporate decision-makers, to help them understand how AI works.

Data privacy is also another emerging problem. The same Asahi report, for instance, raised concerns about privacy issues associated with Abeja’s cameras capturing consumer data for AI analysis. However, the Japanese government currently allows companies to use such data for commercial purposes if it is deleted after analysis. A Parco official was reported as saying that it adheres to this practice.

Okada adds that his company is working on opening up its platform, which was once available solely to corporate customers. It’s now inviting individual engineers, researchers, and data scientists to start projects on the platform. Abeja is also looking to link up with research institutions in Singapore to collaborate on research and development projects.