From Japan to Global Standard AI Foundation Technology: Urata Soft Co., Ltd. - Full-scale Commercialization of Next-generation AI Patent Portfolio Begins

Urata Soft Co., Ltd. (Head office: Nagareyama City, Chiba Prefecture; Representative Director: Tetsuo Urata) announced that in April 2026, it received a patent decision and completed the payment of registration fees for its AI-related patent "Information processing system and information processing method for vectorizing the meaning of words and searching for reasons based on vector similarity" (commonly known as: Why VectorEngine).

This patent was acquired with exceptional speed, receiving an immediate decision (without notification of reasons for refusal) in approximately two months from its application in February 2026.

By combining this patent with the "AI Credit Scoring System" patent (Patent No. 7791512), which the company acquired in December 2025, Urata Soft will officially commence the commercialization of Explainable AI (XAI) in the AI x Finance sector.

- Consecutive Acquisition of Two Patents: Following the AI Credit Scoring Patent (registered December 2025), the Why VectorEngine Patent received a decision in April 2026. Two AI-related patents were acquired in approximately half a year.
- Exceptionally Fast Decision: The Why VectorEngine Patent received an immediate decision approximately two months after its application. This is a result of its clear novelty and inventiveness.
- New Service Concept through Combination: By combining both patents, it is possible to build an "AI Loan Decision and Explanation Generation System" that automatically generates AI loan decisions and their rationales in natural language. This system complies with explainability regulations such as ECOA and FCRA in the US.
- Publication of Technical Explanation and POC Implementation: Technical details of the invention, application examples, and POC implementation sample code are also available on note and GitHub.

note (Technical Explanation): https://note.com/nextgen_urata/n/n1177b5787dc7
Qiita (Technical Article): https://qiita.com/UrataSoft

GitHub (Patent POC Program Publication): https://github.com/NextGenAI-corder

Books (Related Technology): https://amzn.to/4mObVjy

YouTube (AI Credit Scoring Presentation Video): https://youtu.be/8rRYxLN8c1Y

[Patent Overview]

■ Patent No. 7791512 "Credit Scoring System, Credit Scoring Method, and Model Generation System" (Registered December 16, 2025)
The technical core is a system using the LightGBM analysis algorithm in machine learning.
It integrates attribute data and dynamic data, enabling highly accurate delinquency risk prediction with rationale explanation functions using SHAP values. Decision accuracy can be quantitatively proven by Accuracy (correct answer rate) and AUC (probability of successful delinquency prediction).

■ Patent Application 2026-22802 "Information processing system and information processing method for vectorizing the meaning of words and searching for reasons based on vector similarity" (Commonly known as: Why VectorEngine, Patent granted, registration fees paid)
This technology vectorizes the meaning of sentences to search for similar cases, not just conventional keyword searches, and automatically extracts the reasons for similarity.
It can be utilized as core technology for advanced RAG (Retrieval Augmented Generation), LLM hallucination countermeasures, and Explainable AI (XAI).

[Applicable Fields]

While the primary application is in the financial sector (credit decision, loan screening, compliance), other applicable fields are as follows:

- Finance/Credit: Automatic extraction of decision reasons from past similar cases (loan decision and explanation generation in combination with credit scoring).
- Medical: Diagnosis support by searching for similar cases, presentation of rationale for treatment policies.
- Legal: Search for similar precedents/contract clauses, extraction of rationale for judgment prediction.
- SNS Operations: Early warning by detecting similarity with past flame incidents, pre-posting risk assessment.
- Customer Support: Instant presentation of similar cases and solutions from past inquiries.
- HR/Recruitment: Explanation of reasons for aptitude assessment and resignation risk prediction.
- E-commerce/Recommendation: Automatic explanation of "why this product was recommended."
- LLM Hallucination Countermeasures: Tracing output rationale from similar documents (advanced RAG).

[Recruitment for Licensees/Assignees]

Urata Soft Co., Ltd. is widely seeking negotiations for license agreements (preferred) or patent assignments (possible depending on conditions) for the two patents mentioned above, from the following industries:

[Target Industries] Banks, credit unions, consumer finance companies, credit card/loan companies, FinTech companies, medical institutions/medical IT vendors, law firms/legal tech companies, SNS operating companies, customer support-related companies, and other companies utilizing semantic search technology.

Portfolio proposals combining both patents, or discussions on individual cases, are also possible.

[Future Developments]

Urata Soft Co., Ltd., mainly focusing on the AI x Finance sector, plans to sequentially file multiple new inventions, including those currently being prepared for application.

The company is also considering US patent applications, and patent...