Domestic AI Agent Trends (May 23, 2026 Issue)
Update Date: 2026/5/23
Executive Summary
The domestic AI agent trends as of May 22, 2026, show a significant expansion in implementation areas, ranging from financial institution phone reception, sales quoting, and welfare administration to Excel business application conversion, HR analysis, store development, R&D hypothesis generation, contact center operations, security training, and game production. Real-world effects such as automation rates, time savings, and lead time reductions have been demonstrated in cases like Yokohama Bank's certificate issuance reception and NSSOL's quote creation. AI utilization is moving beyond simple consultation to a stage where it completes tasks by linking with existing systems, forms, phone calls, PDFs, and internal data. As cases involving personal information and pre-approved menus increase, balancing convenience with governance is becoming the key to market expansion.
*The content of the created article was input into Gamma to automatically generate slides. Please take a look here if you find slides easier to view.
1️⃣ Yokohama Bank automates phone reception for loan certificate issuance with an AI agent-type voice bot
📎 Source: Mobilus Corporation Press Release
Yokohama Bank has introduced an AI agent-type voice bot using Mobilus's "MOBI VOICE" for a service that automatically accepts applications for loan-related certificate issuance over the phone. For some certificates, it has achieved approximately 90% completion of reception without the need for a callback from an operator, and it expects to reduce annual phone handling and mailing work by approximately 445 hours. The shipping period, which previously took from several days to a week, can now be shortened to as little as the same day through verbal reception operations, making this a notable real-world application of a generative AI workflow-type voice bot in a financial institution.
2️⃣ NSSOL puts quote creation agent into production operation with Agentforce x SVF Cloud
📎 Source: WingArc1st Inc. Press Release
Nippon Steel Solutions (NSSOL) has built a "Quote Creation Agent" that links Salesforce's "Agentforce" with WingArc1st's "SVF Cloud for Salesforce" to automate quote creation tasks. Previously, it took 30 minutes to 1 hour per request and at least 1-2 business days to send, but they have established a mechanism where customers can automatically issue quotes just by selecting necessary items from an AI chat on a portal. By limiting the scope of automatic issuance to pre-approved menus and visualizing processing details, they are promoting efficiency in sales administration while suppressing the risk of erroneous issuance.
3️⃣ Kurashitech launches "Fukushi AI Ope," an AI agent administrative agency service for disability welfare providers
📎 Source: Kurashitech Co., Ltd. Press Release
Kurashitech has launched "Fukushi AI Ope," an administrative agency service utilizing AI agents for disability welfare service providers. The service covers document creation based on service provision records, creation and transmission of billing data to the National Health Insurance Federation, limit management, handling of returns and errors, mailing agency for plans and reports, and even support for complex shift creation. It can be introduced based on existing billing software and chat tools, and by providing communication encryption and access control, it is being deployed as an AI BPO for welfare sites that handle personal and sensitive information.
4️⃣ Digeon officially releases "ENSOU App," which converts Excel into business apps operable by AI agents
📎 Source: Digeon Co., Ltd. Press Release
Digeon has officially released "ENSOU App," which automatically builds AI-native business apps just by uploading Excel files. The AI reads column types, headers, and data structures, allowing business data converted into an app to be operated via chat. AI agents execute tasks such as customer registration, sales aggregation, reading invoices and receipts from PDFs, and report creation, and once a procedure is performed, it can be memorized and reused as a "skill." This approach resolves issues remaining in Excel operations such as transcription errors, reliance on specific individuals, search burdens, and version management while leveraging existing field assets.
5️⃣ NEIGHBOR's "DreamCore" integrates Gemini 3.5 Flash into its game production AI agent
📎 Source: NEIGHBOR Co., Ltd. Press Release
The AI game platform "DreamCore" operated by NEIGHBOR has announced a major update that will sequentially incorporate Google's "Gemini 3.5 Flash" into the core processing of its game production AI agent. DreamCore provides an experience where users can generate code, character images, 3D models, and sounds just by giving natural language instructions, and play or share them immediately in a browser. With this integration, they aim to speed up game template generation, support complex game logic such as multiple characters and stage progression, optimize mobile UI, and strengthen multilingual understanding. The scope of application will be rolled out in stages depending on the user and plan.
6️⃣ Stockmark's "Aconnect" adds "Outlier Hypothesis Generation Beta" feature for R&D
📎 Source: Stockmark Inc. Press Release
Stockmark has equipped the technology exploration agent of its AI agent for the manufacturing industry, "Aconnect," with a new feature called "Outlier Hypothesis Generation Beta." Instead of safe hypotheses that are mere extensions of the past, the AI presents surprising hypotheses that include different technical principles, materials from different fields, and non-traditional processes, supporting the expansion of the exploration range in R&D. Since it presents problem-solving ideas based on vast amounts of papers and news and allows for checking comprehensiveness in a logic tree format, it is expected to be used for organizing discussion points before meetings and creating medium- to long-term research themes.
7️⃣ KnowBe4 Japan launches AI defense agent for security training
📎 Source: KnowBe4 Japan K.K. Press Release
KnowBe4 Japan has announced a new plan for its security awareness training product and features related to its AI defense agent, "AIDA." In SAT Advanced, it links with an Orchestration Agent that automates program operations and a Deepfake Training Content Agent that can create deepfake videos reproducing organizational leaders. The new Content Creation Agent generates training packages including training content and comprehension tests based on internal regulations and materials, and supports delivery in 30 languages as well as LMS export.
8️⃣ WHI Holdings incorporates custom AI agent that reads personnel systems into CTM 2.0
📎 Source: WHI Holdings, Inc. Press Release
WHI Holdings has integrated an AI agent into its "COMPANY Talent Management" series that handles everything from HR data analysis to suggesting improvement measures, based on the company's unique personnel systems and organizational characteristics. Initially available in the "Motivation Survey" product, the custom AI agent, which has been trained on evaluation system documents and grade definitions, analyzes free-text responses and engagement scores within the context of the company's systems. It goes beyond general attribute analysis to support the identification of issues by grade and evaluation, and suggests improvement actions such as 1-on-1 meetings or challenging assignments.
9️⃣ Nowcast provides "DataLens Store Development" AI agent to Oizumi Foods
📎 Source: Finatext Holdings, Inc. Press Release
Nowcast, a member of the Finatext Group, has provided its "DataLens Store Development" AI agent to Oizumi Foods, which operates restaurant and fitness businesses. For staff who receive over 600 property listings per month via email or PDF, the AI automatically analyzes the property information and assists in plotting it on maps and creating lists. By combining trade area analysis based on human flow and payment data with generative AI, it enables centralized management of property information, comparison of coverage areas with existing stores, and prioritization of potential store locations. This is a case study of shifting store development operations, which often rely on intuition and experience, to a data-driven approach.
🔟 Mobials "MooA" to sequentially launch customer harassment (kasuhara) monitoring and knowledge agent functions
📎 Source: Mobials, Inc. Press Release
Mobials will begin sequentially providing six new features for its contact center operation support AI, "MooA," starting in May 2026. The "MooA CommNavi" response support dashboard will add generative AI-based call status monitoring, customer harassment (kasuhara) prediction detection, improved speech recognition engines, and a memo template function. "MooA KnowledgeBase" will implement a knowledge agent function that automatically extracts knowledge from interaction logs, as well as search and management functions. This move aims to simultaneously protect operators, formalize response know-how, and improve inquiry resolution speed.
Comprehensive Analysis
From the topics on May 22, 2026, a clear trend emerged: domestic AI agents are shifting from "convenient chat tools" to "execution entities embedded in business processes." In particular, there are notable cases where AI converts existing business assets—such as phone calls, forms, Excel files, PDFs, interaction logs, HR systems, and trade area data—into a format it can handle, thereby replacing or supplementing actual on-site procedures. On the other hand, in fields involving sensitive information or decision-making responsibility, such as finance, welfare, HR, and security, simple automation is insufficient. Companies that can incorporate control over authorization scopes, visualization of processing content, human intervention in exceptional cases, and auditability will be able to continuously increase the results of their AI agent implementations.
Future Points of Interest
AI agents are beginning to handle reception, quoting, and form issuance; the focus will now be on whether they can measure not only time savings but also improvements in customer experience.
In welfare, HR, store development, and R&D, the differentiator will be how industry-specific systems and on-site knowledge are incorporated, rather than the performance of general-purpose models.
Products that leverage existing assets like Excel, PDFs, and interaction logs may spread as a realistic AI adoption route for companies that want to avoid system overhauls.
In the contact center sector, the trend toward integrating customer harassment detection, response quality management, and knowledge extraction—beyond just automated responses—is likely to strengthen.
As the areas where AI is granted execution authority increase, the design of approval scopes, log management, personal information protection, and liability in the event of processing errors will become a competitive advantage in itself.
