Commencement of Joint Development of "M&A Sourcing AI" for Financial Institutions and M&A Advisory Firms

Buffett Code Inc. (Headquarters: Meguro-ku, Tokyo; Representative Director: Tomohiro Fukuda; hereinafter referred to as Buffett Code) and Elith Inc. (Headquarters: Bunkyo-ku, Tokyo; Representative Director: Koki Inoue; hereinafter referred to as Elith) have commenced joint development of "M&A Sourcing AI," an AI agent designed to assist in creating lists of potential acquisition targets for M&A.

This PoC will combine Buffett Code's high-quality corporate database for M&A with Elith's AI agent platform for the financial sector, which has also been adopted by GENIAC4, to provide an AI agent that meets advanced M&A expertise and the security requirements demanded by M&A operations.

4Adopted by GENIAC 4th Term: Ministry of Economy, Trade and Industry (June 4, 2026) https://www.meti.go.jp/press/2026/06/20260604003/20260604003.html

M&A Sourcing AI: A Key Driver for Transforming the Financial Industry's Business Structure

M&A involving domestic companies reached a record high of 5,115 cases in 2025, an 8.8% increase year-on-year (Source: M&A Capital Partners, "M&A Market Size in 2025"). The demand driven by business succession and a shortage of successors is expected to expand further.

Furthermore, it is important to note that M&A sourcing is both a "social issue" and a significant revenue opportunity for financial institutions. In particular, for regional financial institutions affected by the shrinking regional economy, establishing pillars of revenue beyond lending is an urgent challenge. Among these, M&A matching, including business succession support, is an area expected to become a new business pillar due to the stable commission income it can generate.

In this context, regional financial institutions are in an advantageous position. Through daily lending transactions and dialogues with business owners, consultations regarding successor shortages, concerns about business future, growth investment, and capital alliances are first brought to regional financial institutions.

However, there is a challenge in not fully leveraging this advantageous position. Finding optimal buyers requires industry understanding, corporate data to search for potential buyers, extensive research efforts, and knowledge of M&A practices. Many financial institutions are forced to rely on their existing client networks and the experience of their staff, which tends to narrow the scope of buyer searches.

As a result, many cases end with referrals to external M&A intermediaries. This leads to the outflow of potential commission income opportunities to external parties.

We believe this situation is not unique to financial institutions but is an industrial structural issue where human-centric M&A support models are reaching their limits amidst the simultaneous decline of regional companies, staff shortages, lack of specialized personnel, and fragmented information.

Therefore, there is a need for a system that can broadly, quickly, and reproducibly search for buyers and sellers while leveraging the greatest asset of financial institutions: customer touchpoints. Additionally, M&A requires high levels of confidentiality and information control, making the introduction of general-purpose AI a high hurdle for financial institutions, which are subject to strict security and compliance standards.

This joint development will provide "M&A Sourcing AI," which combines Buffett Code's high-quality corporate database for M&A with Elith's AI agent platform that meets the security requirements of financial institutions, enabling the search for M&A buyers and sellers without relying on human resources and experience.

Why Have M&A Challenges Not Been Resolved?

M&A Sourcing AI requires the fulfillment of the following four conditions, but all have been difficult to achieve simultaneously:

Requirement 1: Comprehensive Corporate Data is Required

Creating lists of potential seller and buyer companies, while seemingly simple, is actually a highly challenging task.

First, the sourcing process of listing potential companies that can create synergy with one's own company and then narrowing them down by conditions increases the probability of M&A success. However, web searches can only gather information on listed companies, and even paid corporate databases often do not offer services that enable comprehensive listing of companies, including unlisted ones.

Requirement 2: Screening from Various Perspectives is Possible

In addition to comprehensiveness including unlisted companies, a candidate list cannot be created without the ability to screen these companies based on factors such as company size, growth potential, and business complementarity. This requires the inclusion of diverse information items such as company business content, financial data, and shareholder composition. Furthermore, for unlisted companies whose financial statements are not disclosed, "proxy indicators" that capture scale and growth through alternative metrics are necessary for decision-making.

Requirement 3: Lists Can Be Created Using Natural Language

Even with a comprehensive corporate database containing diverse items, specialized knowledge is required to list optimal M&A candidates. This is because it is necessary to verbalize the desired acquisition and translate it into searchable items.

However, in a situation with a shortage of personnel where anyone is expected to be able to perform the task, a system that assumes such expertise is difficult to utilize. "I want to find companies that would likely acquire this company," or "I want to identify potential capital alliance partners in this business domain." An AI agent that can extract candidates and provide reasons for selection simply by posing such ambiguous questions is the key to overcoming the lack of expertise.

Requirement 4: Security

M&A is an area that demands high levels of confidentiality and information control. Without risk controls for issues like hallucinations and prompt injection, it cannot be used in the mission-critical domain of M&A. Moreover, M&A requires advanced information control. If risks of information leakage, such as confidential data being used for training generative AI models or project members outside of a specific project accessing important files via an AI agent, cannot be controlled, it cannot be utilized in the core operations of financial institutions.

Any one of these four requirements being missing prevents practical application. A comprehensive corporate database optimized for M&A use and an AI agent platform that meets the security requirements of financial institutions. This was made possible only by the collaboration of players with these two distinct areas of expertise.

Breaking Through This Barrier with Buffett Code × Elith

Buffett Code and Elith will realize "M&A Sourcing AI" by combining the strengths of both companies.

Buffett Code will provide a comprehensive corporate database and detailed information items built by mobilizing years of M&A expertise. Elith will provide the AI expertise and implementation capabilities to make this accessible to anyone in natural language, along with the security required by financial institutions. Thus, the ideal M&A Sourcing AI, which has been unattainable until now, is realized here.

Buffett Code's Strengths

Comprehensive Corporate Database Capable of M&A Practice

Approximately 1.6 million companies in Japan, including listed and unlisted, are structured in a way that facilitates the identification of M&A targets. It provides a corporate data foundation that can be used not only for simple company listings but also for initial screening and organizing proposal hypotheses.

Abundant Information Items for Acquisition Hypotheses and Screening In addition to 2,393 types of industry classifications, screening from an M&A perspective is possible based on business content, size, growth potential, financial information, shareholder information, funding history, and employee number trends, enabling searches for "companies with complementary products," "companies in adjacent markets," "companies with the same customer base," and "companies located upstream or downstream."

Proxy Indicators to Capture Changes in Unlisted Companies For unlisted companies whose financial statements are not disclosed, alternative indicators such as monthly employee number trends are used to capture signals of company size, growth, or contraction. This supports candidate searches from a broad pool, not limited to existing clients or listed companies.

Data Accumulation Optimized for AI Utilization

AI cannot make correct judgments with data that has merely been collected. Buffett Code has refined the quality of its database through repeated data shaping, such as mapping account items and absorbing accounting differences, accumulated over a long period. This quality forms the foundation for the accuracy of the AI agent's responses.

Elith's Strengths

Expertise in AI Security / AI Guardrails Addresses risks such as AI malfunction, information leakage, prompt injection, and regulatory compliance. Responsible for implementing AI agents that can be used with confidence by regulated industries, including financial institutions.

AI Technology Centered on the Quality Evaluation Platform "GENFLUX" Provides technology to evaluate the response quality of generative AI and detect/improve risks related to misinformation and regulatory compliance. Based on expertise in AI Safety / Trust & Safety, it supports safe AI utilization with the premise that "AI makes mistakes."

Fast Implementation and Improvement Capabilities with FDE Model

Possesses development capabilities to rapidly cycle through business requirement concretization, prototype development, and improvement, working alongside clients' operational sites. Supports implementation tailored to the unique business flows and security requirements of each financial institution.

High Trust in Technical Capabilities

Recognized for its R&D capabilities and social implementation in AI technology and AI security, including adoption by METI/NEDO's "GENIAC" and reaching the finals of Japan's largest startup conference, "IVS 2026 LAUNCHPAD."

Changes Brought About by "M&A Sourcing AI" and Beyond

M&A Sourcing AI will bring about the following transformations in M&A advisory fields, including regional financial institutions, PE firms, and investment banks.

AI will assist in the discovery of buyers and sellers, the biggest challenge in M&A. For example, to find a buyer for an automotive glass manufacturing company struggling with a succession shortage, simply posing an abstract condition like "Find automotive parts manufacturers growing in recent years in Northern Kanto" will lead to the extraction of candidate companies by combining multiple signals, along with reasons for selection. The "discovery of deals," which previously relied on personal networks and intuition, can now be performed quantitatively and safely through structured data and reasoning AI.

In the medium term, we aim to expand the scope of support beyond sourcing. We aim to reach a state where AI can provide end-to-end support for the M&A process, from initial brainstorming on M&A strategy formulation, to creating and shortlisting long lists, valuation, and due diligence (DD).

[Company Overview]

Buffett Code Inc. Buffett Code is a corporate analysis database that enhances the accuracy and efficiency of information gathering and analysis for those involved in corporate planning, finance, M&A, and new business development. It contains financial information and management indicators for approximately 1.6 million companies (as of February 2026), including listed and unlisted Japanese companies, and provides an analysis environment that allows for everything from listing target companies to comparative analysis, equipped with 2,393 industry classifications and 247 screening conditions. By organizing scattered corporate information and delivering it in an usable format, it contributes to improving the quality and speed of decision-making.

<Company Overview> Company Name: Buffett Code Inc. Location: Aobadai 1-14-2-301, Meguro-ku, Tokyo Representative: Tomohiro Fukuda, Representative Director Business Activities: Development and operation of the corporate analysis platform "Buffett Code" URL: https://www.buffett-code.com/

Elith Inc. Elith Inc. is a tech company that co-creates optimal solutions with AI by discovering challenges alongside its clients. For a wide range of industries including manufacturing, finance, and healthcare, it provides integrated solutions from consulting to the development of generative AI, LLM, and image AI, as well as AI education and advisory services, centered on AI solution provision.

Furthermore, it focuses on AI Safety, which is increasingly important for the social implementation of generative AI, and supports the construction of AI infrastructure that can be used with confidence through its AI safety platform "GENFLUX," which checks the response quality and operation of generative AI in real-time and automatically provides improvement suggestions.

Company Name: Elith Inc.

Representative: Koki Inoue, Representative Director CEO & CTO

Head Office Location: Frontier Hongo I 6-A, 2-27-17 Hongo, Bunkyo-ku, Tokyo

Business Activities: Research, development, design, planning, education, sales, maintenance, and consulting services related to AI

Company Overview URL: https://elith.ai

[Inquiries Regarding This Matter / PoC]

◆Inquiries Regarding This Matter / Consultation for PoC https://business.buffett-code.com/contact?utm_source=prtimes