\ Innovation trend commentary /

In this series, we will introduce the themes and ventures that were taken up in each morning pitch, and send out information that contributes to Japanese innovation.

This time, we will be holding a special event on February 22nd.Generative AI special feature"is.

Global fundraising volumes set to rise sharply from 2023

Generative AI is a general term for AI that generates text, images, source code, etc. Unlike conventional AI that only learns and analyzes data, generative AI can understand and learn natural language and other diverse information to handle a wide range of tasks and return natural content as if it were generated by a human. Until the year before last, the main focus was on text and images, but from 2023 it will be able to handle a variety of creations, including images, source code, and videos, such as OpenAI's Sora.

Looking at the amount of global funding in the field of generative AI, it has risen sharply since the beginning of 2023. It can be said that this is an area that has attracted attention amid a slump in the market for startups. However, about 70% of the total amount raised is accounted for by large-scale funding from OpenAI and US Inflection AI, and about 80% of the market is made up of early-stage startups, and companies are still in the process of exploring in which areas generative AI will be accepted.

Looking at the areas, AI assistants are by far the most popular. In other sectors that produce unicorns (private companies with a corporate value of over $1 billion), funds are being funneled into areas where the benefits of introducing them are easy to see, such as improving internal productivity, streamlining marketing, sales, and customer service.

Adoption is accelerating in areas related to improving employee and organizational productivity

The needs for companies to utilize generative AI can be divided into four cases. The most advanced adoption is in areas related to improving employee and organizational productivity. The second is where generative AI is incorporated into business systems to automate and improve quality. The third is in the area of customer contact, where we feel that the number of cases using avatar videos in advertisements and to create experiences has increased since last year. The fourth is the development of custom-made large-scale language models (LLMs) in our own domain, where LLMs that incorporate industry information such as construction, finance, and pharmaceuticals are used in various new businesses and development around DX.

Increasing collaboration with major companies

When large companies use and introduce generative AI, they are increasingly partnering with startups. A subsidiary of a major railroad company that handles customer service functions has partnered with ELYZA, a startup from the University of Tokyo, to semi-automate customer support operations. Generative AI is not only used to create responses to as many as 6,000 inquiries per day. The company has established a business process in which AI is used to create summaries of requests for internal use, and humans simply check them. A major general contractor is working with Akari, another startup from the University of Tokyo, to make effective use of BIM (Building Information Modeling), which links information about building materials with 3D models. Specifically, the company is developing a tool specialized for the construction industry that allows users to ask in natural language questions such as, "How much did the facility renovation carried out three years ago cost?"

Understanding the regulatory situation in each country and region is essential

If you are developing a new business that utilizes generative AI and are considering global expansion, it is important to understand the regulatory situation in each country. Europe has strong AI regulatory laws, requiring companies that provide generative AI to clearly indicate that it is made by AI, and full enforcement will begin in 2026. China requires disclosure of the algorithms used in AI, effectively excluding foreign-made generative AI such as ChatGPT. Japan has not yet established solid regulations, but discussions on regulations are underway in the wake of the G7 Hiroshima Summit in May 2023.

Although overseas companies are far ahead of the domestic AI startup market, startups are also emerging in Japan, focusing on AI model development and application layers. There are also many cases of overseas startups setting up bases in Japan to promote their businesses, and three of the five companies introduced here are from overseas.

Japanese-specialized LLM receives high marks in Japanese language assessment criteria (Deeppreneur Co., Ltd.)

Deepreneur is a group of engineers and researchers, mainly consisting of members from the University of Tokyo's Matsuo Laboratory. They develop digital transformation solutions for corporations using generative AI, including large-scale language models (LLMs). Their strengths lie in their unique generative AI technology and their track record of joint research with large companies. Their Japanese-specific LLMs boast performance that overwhelms ChatGPT3.5 and other systems in the Japanese evaluation index "JGLUE." In the future, they plan to further popularize their solutions by utilizing flexible architectures such as additional learning.

Customer support services via chat with AI generation (wevnal Co., Ltd.)

Wevnal offers "BOTCHAN AI," a customer support service that can be conducted through chat with generative AI. As opposed to the marketing funnel, where the number of prospective customers is gradually narrowed down to those who make a purchase, wevnal's service improves CVR (conversion rate) and maximizes LTV (customer lifetime value) by providing comprehensive customer service throughout the entire process. As it is a performance-based payment system, the initial implementation risk is low, and the service has been implemented in over 600 companies to date. In the future, wevnal will strengthen its collaboration with Microsoft to enable 24/7 automated responses and personalized customer service tailored to the user's interests.

Processing of over 50 personal identification codes into 52 languages including Japanese (Private AI Inc.)

Private AI Inc., a Canadian venture, owns technology to identify and anonymize personal identification codes using machine learning models that utilize world-class algorithms. It is capable of processing over 50 personal identification codes in 52 languages, including Japanese, and has already been adopted by government agencies and large companies in North America and Europe. Since it can also be adapted to data types unique to Japan, such as My Number and address notation, the company plans to actively expand into the Japanese market in the future, not only through direct sales but also through partnership models.

Intuitive, no-code generative AIAllganize Japan Co., Ltd.)

Allganize Japan, a Silicon Valley-based global AI solutions company, offers a service that allows users to intuitively utilize generative AI without coding, to solve the problem of not knowing how to link their company's data with generative AI. It supports over 100 use cases, from general to specialized, and also provides functions that allow users to streamline specific situations without coding. The big difference between Allganize Japan and similar tools is that it pursues and realizes a UX that anyone can use.

A free, no-code AI tool that doesn't require specialized knowledge (Leeton Technologies Japan Co., Ltd.)

Leeton Technologies Japan, a Korean startup, offers Leeton Studio, a no-code AI tool that is free to use and does not require specialized knowledge. Using this service, you can efficiently create AI agents that work with your company's data systems and perform the necessary tasks. Generative AI technology can be applied to improve productivity at low cost. The first service was launched in October 2022, one month earlier than ChatGPT. Since then, it has achieved a cumulative user base of 2 million in one year and is rapidly expanding in the Japanese market.

Japan is facing a declining birthrate and an aging population, and in order to achieve sustainable economic growth, improving productivity using generative AI is an urgent issue. Expectations are high for the further rise of leading Japanese startups in the field of generative AI.

▼Theme Leader Profile

Deloitte Tohmatsu Venture Support Co., Ltd.

Business Produce Department

Kensuke Machida

Graduated from the Faculty of Engineering at the University of Tokyo. Worked as a product manager for a total of three products at an IT startup, and engaged in the growth of BtoC media and the launch of BtoB SaaS. In his current position, he is mainly responsible for new business development for major companies, including the field of generative AI.

-For those who want to catch up on innovation trends on a regular basis-

—————————————————————————————————————————————-

In Morning Pitch, we will explain the outline of each theme as described above.

We deliver materials and videos only to paying members.

　Explanatory material ...Morning Pitch paid member

Explainer video ...Morning Pitch Innovation Community (MPIC) member

—————————————————————————————————————————————-