Domestic AI Agent Trends (June 12, 2026 Issue)
Update Date: 2026/6/12
Executive Summary
June 11, 2026, marked a shift where the implementation of AI agents expanded cross-sectorally into payments, professional services, marketing, analytics, regional media, support, and security. Visa and Mastercard introduced authentication, authorization, and payment infrastructure to support inter-agent transactions, while in the business domain, Sharoume, NoimosAI, MANA Studio, Agentforce, and Tableau are advancing integration into specialized tasks. In the domestic market, interest is shifting from standalone chatbots to agents that receive, execute, and report on tasks, and further toward infrastructure for secure control.
*The created article content was input into Gamma to automatically generate slides. If you find slides easier to view, please take a look here.
1️⃣ Visa and Mastercard announce payment infrastructure to support transactions between AI agents
📎 Source: Visa Worldwide Japan Press Release, Mastercard Press Release
At the Visa Payments Forum 2026, Visa announced "Visa Intelligent Commerce," which supports AI agents in securely discovering, initiating, and completing transactions on behalf of consumers and businesses. The company combined agent scoring to evaluate AI agents, an agentic directory to list trusted agents, and large transaction models trained on vast payment data, while also highlighting collaborations with OpenAI and the expanded use of tokens and stablecoins. Additionally, Mastercard announced "Agent Pay for Machines," which supports high-frequency, low-value payments between AI agents and machines through authorization management, spending limits, and payments via cards, accounts, and stablecoins.
2️⃣ MK System to equip "Sharoume" with the AI agent function "AI Sharoume"
📎 Source: MK System Co., Ltd. Press Release
MK System will launch "AI Sharoume," an AI agent function that integrates with the labor and social security attorney business support cloud "Sharoume," scheduled for July 23. It supports tasks such as "reading, searching, inputting, and confirming," including entry/exit procedures, checking calculation basis notifications and monthly change notifications, reading official documents and return contents, and detecting errors before transmission. Assuming final confirmation and approval by humans, it assists with daily operations through five agents: procedures, payroll, social insurance, employment insurance, and labor insurance.
3️⃣ NoimosAI launches autonomous AI marketing team for Japan
📎 Source: Agos Labs Technologies LTD. Press Release
Agos Labs Technologies has officially launched "NoimosAI," an autonomous AI marketing team, for the Japanese market. Multiple AI agents autonomously and collaboratively execute tasks ranging from competitor analysis, SEO/GEO strategies, content creation, and SNS management to performance measurement and PR, providing reports and seeking approvals via email or Slack. By integrating with external data sources like Semrush and Apollo and existing applications, it provides small and medium-sized enterprises and individuals with the "experience of having a marketing team," backed by achievements such as over 10,000 monthly visits, 10x ROI, and over 20 million total monthly reach.
4️⃣ Givery's "MANA Studio" adds same-day support for Claude Fable 5
📎 Source: Givery, Inc. Press Release, Anthropic Official Announcement
Givery has begun supporting Anthropic's latest model, "Claude Fable 5," in its corporate generative AI and AI agent utilization platform, "MANA Studio." MANA Studio allows users to switch between multiple LLMs and build AI agents specialized for specific tasks, called "MANA Buddy," without coding. While Fable 5 excels at long-duration and complex tasks, it features safeguards that fall back to Claude Opus 4.8 if its classifier detects risks in high-stakes areas such as cybersecurity, biology, chemistry, or model distillation, a design that provides both high performance and secure control.
5️⃣ Gunosy develops AI agent that handles everything from regional information collection to article distribution
📎 Source: Gunosy Inc. Press Release
Gunosy has developed and launched a "hyper-local article generation system" on May 21 that utilizes AI agents to streamline everything from acquiring regional information to creating and distributing articles. This system collects and organizes crime prevention and safety information, municipal announcements, local events, and store information issued by government agencies and local municipalities that have granted distribution permission, and structures them as articles tailored to the user's living area. A relatively high open rate has been confirmed for push notifications, and the company plans to expand the target regions and the scope of distributed information while increasing collaboration partners.
6️⃣ Recho and Q&A partner on AI voice agent for technical support
📎 Source: Recho, Inc. Press Release
Recho has entered into a strategic partnership with Q&A, which operates an ICT technical support business, to develop and provide AI voice agents specialized in the technical support domain. To handle highly difficult interactions that require everything from clarifying situations based on vague customer reports to estimating causes and determining the necessity of repairs or on-site visits, they are combining Recho's proprietary speech recognition, speech synthesis, and dialogue control technology with Q&A's on-site expertise. Collaboration has already begun at a major home appliance manufacturer's contact center, and both companies aim to deploy a model in the enterprise domain that optimally combines AI Voice Agents with human interaction.
7️⃣ marutto launches "Marutto AI Agent" for small and medium-sized enterprises
📎 Source: marutto Inc. Press Release
marutto has launched "Marutto AI Agent," an AI agent service for small and medium-sized enterprises. Targeting companies where general-purpose generative AI tools have not taken root, it provides a comprehensive package including business inventory workshops, design and construction of customer-specific agents, and monthly subscription-based operational support. Aiming to be an "employee-type" agent that receives, executes, and reports tasks rather than just a "dictionary" type for Q&A, its strength lies in operational design expertise—including quality checks, security, and internal rules—cultivated through in-house practice in back-office operations, financial reporting, and meeting minute task aggregation.
8️⃣ Frogwell supports the introduction of Agentforce similarity search for Fischer Instruments
📎 Source: Frogwell Inc. Press Release
Frogwell has provided construction and implementation support for a "Past Similar Record Search Function" utilizing Salesforce's AI agent platform, "Agentforce," for the Japanese subsidiary of measurement instrument manufacturer Fischer Instrumentation. The system automatically groups product name variations through proprietary preprocessing and uses AI to semantically search and list past records with "30% or higher similarity," taking the context of defect symptoms into account. This makes it easier to access past resolution procedures that could not be found through exact-match searches, serving as an agent case study that supports the elimination of reliance on individual skills and the transfer of technical expertise, allowing even junior staff to handle initial responses without depending on the veteran's rule of thumb.
9️⃣ Salesforce to Deploy Tableau's Agentic Analytics Platform in the Japanese Market
📎 Source: Salesforce Official Announcement
At DataFam Tokyo, Salesforce announced the deployment of Tableau's next-generation AI agent-based analytics platform, "Agentic Analytics Platform," in the Japanese market. They presented six pillars, including a knowledge engine that allows AI to understand company-specific sales definitions and business rules, "Tableau Agent" for conversational analysis in natural language, a decision-making engine that connects analysis results to actions such as case creation in Salesforce, and headless functions that deliver insights to external interfaces like Slack and Teams via MCP servers. Auto Knowledge Graph is scheduled for general domestic release in July, conversational analysis in late 2026, and the release date for the Agentic Analytics Command Center remains undecided.
🔟 MBSD and SMBC Cyber Front to Develop Domestic Guardrail Products for AI Agents
📎 Source: Mitsui Bussan Secure Directions, Inc. Press Release
MBSD and SMBC Cyber Front announced the start of development for domestic security products designed to safely and securely utilize AI agents and generative AI. The concept involves inspecting and controlling dangerous AI agent behaviors, such as detecting the use of unauthorized AI services (Shadow AI), the external transmission of confidential information, and the execution of dangerous commands, as well as detecting attacks that are difficult to identify with conventional rule-based systems, such as multi-turn memory poisoning attacks, through multi-layered control. The system features a choice of gateway or proxy configurations, a combination of real-time and post-event inspection, and centralized management of action logs, with a PoC planned for the first half of 2026 and a beta release in the second half.
Comprehensive Analysis
The key feature observed from the topics on June 11, 2026, was that AI agents have begun to be positioned not as "convenient generative AI functions," but as execution entities that integrate into business processes across industries. In payments, identity, authorization, and expenditure control have become the focus, while in labor and social security consulting, marketing, analytics, and support, the source of value lies in connectivity with existing SaaS and on-site data. While standardization by major platforms and hands-on implementation by field-oriented services are running in parallel, new risks such as Shadow AI and memory poisoning are emerging. Moving forward, auditability, human approval, and division of responsibility will become conditions for widespread adoption, in addition to performance.
Future Points of Interest
AI agent payments are likely to shift their competitive axis toward building a foundation of trust that proves "who authorized what," rather than just automated payments.
In specialized fields such as labor and social security consulting, support, and analytics, specialized agents that are deeply connected to business data and approval flows will have an advantage over general-purpose AI.
In the market for small and medium-sized enterprises, demand for hands-on services that include not only tool provision but also business inventory, design, and operational improvement is likely to strengthen.
In areas where AI handles everything from information gathering to distribution and response, such as hyper-local articles and voice interaction, quality control and accountability will become critical.
The emergence of domestic guardrail products is seen as a sign that AI agent utilization is moving from the verification stage to full-scale operation premised on control and auditing.
