Domestic AI Agent Trends (June 27, 2026 Issue)
Update Date: 2026/6/27
Executive Summary
The domestic AI agent trends for June 26, 2026, saw a succession of task-specific implementations ranging from log analysis, medical appointments, and technical knowledge transfer in manufacturing, to sales support, labor management, internal search, support for government and politicians, wearables, source code analysis, chat-based knowledge conversion, and economic evaluation benchmarking. The common thread is that they go beyond simple text generation to connect with existing systems and business data, taking on tasks such as judgment, execution, summarization, and the creation of next actions. In particular, designs that incorporate safety and accountability boundaries—such as BYOC (Bring Your Own Cloud), permission integration, explainability, and human approval—are prominent.
*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 BTM launches autonomous log analysis AI agent 'TracisAI' on AWS Marketplace
📎 Source: BTM Inc. Press Release
BTM Inc. has launched its autonomous log analysis AI agent 'TracisAI' on the AWS Marketplace. Starting from error detection or inquiry text, the AI automatically extracts relevant logs, generates and executes SQL, identifies causes, and provides summaries. It has demonstrated a track record of reducing investigation times that previously took several hours to about 10 minutes, a reduction of up to 95%. Since inquiries can be made in Japanese via Slack, sales and support staff can obtain primary answers without going through engineers. A key feature is its design, which adopts a BYOC approach built and operated within the customer's AWS environment, suppressing the risk of log or personal information leakage.
2️⃣ Omi Japan announces AI agent infrastructure for medical care and appointment change agent 'MediDial'
📎 Source: Omi Japan Inc. Press Release
Omi Japan has announced an AI agent development framework for medical institutions and is rolling out 'MediDial' as its first product to support appointment change tasks. The design allows the AI to call necessary functions according to the situation and proceed with tasks while judging the processing order, based on the prerequisites of security, explainability, existing system integration, and response performance, which are challenges in medical settings. MediDial handles appointment requests via chat and voice, branching to full automation for low-risk tasks, staff approval for medium-risk, and human intervention for high-risk. This serves as a case study for clarifying the division of responsibility between humans and AI in the medical field.
3️⃣ Dentsu Soken to equip iQUAVIS with 'Technical Breakdown AI Agent'
📎 Source: Dentsu Soken Inc. Press Release
Dentsu Soken has announced that it will equip its product development support solution for the manufacturing industry, 'iQUAVIS,' with a proprietary 'Technical Breakdown AI Agent,' with plans to provide an automatic technical information structuring function in January 2027. The AI organizes and structures technical information such as requirements, functions, parts, and manufacturing processes, converting personalized experience and tacit knowledge into a state where it can be accumulated and reused as explicit knowledge. By combining iQUAVIS, which is introduced in over 220 companies in Japan and overseas, with the corporate generative AI platform 'Know Narrator,' it has shown a direction for vertical agents for the manufacturing industry oriented toward 'Technical Succession DX,' supporting the structuring of technical information including images and charts, impact analysis during design changes, and technical succession and quality improvement.
4️⃣ Comture launches 'Sales Support AI Template' with Salesforce integration
📎 Source: Comture Corporation Press Release
Comture has launched a 'Sales Support AI Template' that combines Salesforce with generative AI. By inputting sales conversation audio, meeting memos, and materials, it provides end-to-end support including generating drafts of meeting minutes and reports on Salesforce, quantitative and qualitative scoring of conversation content, drafting thank-you emails, turning meeting information into knowledge, reflecting customer requests and sales status in CRM fields, and even SWOT analysis utilizing public information such as EDINET and internal documents. It starts small as a 3-month, 1.5 million yen verification pack, with a vision to expand to company-wide AI agent utilization beyond sales through data infrastructure development utilizing Informatica in the future.
5️⃣ Seven Sense Marketing releases labor management use case with PC logs x Claude MCP integration
📎 Source: Seven Sense Marketing Inc. Press Release
Seven Sense Marketing has released a new use case that integrates its PC log management cloud 'Mieru Cloud® Log' with the AI agent 'Claude' via MCP. It visualizes hidden overtime that cannot be grasped by attendance data alone from objective data such as PC operation logs and screenshots, and by simply asking the AI, it performs tasks from extracting overtime workers to factor analysis and suggesting business efficiency improvements. It demonstrated a practical example of seamlessly automating 'detection -> analysis -> next action creation' in labor management, with the AI automatically creating analysis reports at the beginning of the month and even generating draft warning emails for supervisors of relevant departments if there are issues.
6️⃣ Knowledge Sense 'ChatSense' enables AI search of Teams data with permission integration
📎 Source: Knowledge Sense Inc. Press Release
Knowledge Sense has released a new feature for its corporate generative AI agent 'ChatSense' that allows users to search Microsoft Teams data from the standard chat input field. Users can search channel posts and meeting memos within the scope of their own Microsoft account permissions, allowing them to safely reference only the necessary internal knowledge without passing all company-wide Teams information to the AI. As the integration targets for corporate AI, used by over 500 companies including those listed on the Tokyo Stock Exchange Prime Market, expand, permission integration and the suppression of over-exposure have become important design points for internal search by AI agents.
7️⃣ First Innovation to begin accepting applications for 4 models of dedicated AI agent OS 'Lumina'
📎 Source: First Innovation Inc. Press Release
First Innovation has announced that it will begin accepting applications from July 1st for four models of 'Lumina,' a dedicated AI agent OS that turns the knowledge of companies, executives, local governments, and politicians into AI: 'Lumina Corporate,' 'Lumina President,' 'Lumina Government,' and 'Lumina Political.' Based on business card information, organizational materials, websites, policy information, activity reports, and past communications, it generates dedicated AI agents to support text creation, SNS posting, document creation, public relations, inquiry handling, policy and measure information organization, activity reporting, and executive message creation. It aims to develop an AI utilization environment that contributes to increasing the growth speed of Japan as a whole by deploying the knowledge gained from supporting the creation of over 300 AI agents for its own company and other companies/organizations as domain-specific packages.
8️⃣ UNTITLED to launch AI wearable 'AIomi' in Japan
📎 Source: UNTITLED Inc. Press Release
UNTITLED announced that it will launch the AI wearable "AIomi" in Japan starting June 26, 2026, via the crowdfunding service "Makuake." This is the Japanese market version of the "omi" AI wearable developed by the U.S.-based Based Hardware Inc. It is positioned as a "personal AI memory" device that records, summarizes, and structures conversations simply by being worn, allowing users to search for necessary information using natural language. Furthermore, by integrating with over 1,000 applications, it functions as an AI agent wearable that supports task execution while connecting conversation content to existing systems such as CRM and task management. Backed by a track record of over 300,000 users worldwide, it aims to bridge the gap between human memory and business actions in meetings, negotiations, interviews, and learning.
9️⃣ TechMatrix "Understand 7.2" provides MCP functionality for AI agents
📎 Source: TechMatrix Corporation News Release
TechMatrix began sales of the Japanese version of the source code analysis tool "Understand 7.2" on June 25, officially providing generative AI integration and MCP functionality. AI agents can now access the analysis database (.und file) via the Understand API to directly retrieve entities, references, dependencies, metrics, and control flow information. When developers query analysis results in natural language, the AI determines and retrieves the necessary information and automatically generates explanations and improvement suggestions, making it easier to build AI coding agents based on high-precision analysis data.
🔟 Fragment officially releases chat AI agent "NEO"
📎 Source: Fragment Inc. Press Release
Fragment has officially released "NEO," a chat AI agent that transforms team chat history into corporate knowledge assets. Rather than simple chat support, it understands and remembers conversations, structuring requirements, decisions, tasks, and the background of decision-making to accumulate them as project knowledge. After free registration on the official website, users can accumulate conversations and documents in workspace or project-specific threads. It has become a domestic AI agent official version that utilizes communication that tends to flow away and disappear as reusable long-term memory for requirement definition, sales reviews, recruitment/HR, and knowledge transfer in manufacturing settings.
1️⃣1️⃣ KPMG Azsa and Sakana AI announce multi-agent economic evaluation "CoffeeBench"
📎 Source: KPMG Azsa LLC Press Release
KPMG Azsa has concluded a joint research agreement on auditing with Sakana AI and, as its first achievement, announced "CoffeeBench," a benchmark for long-term simulation of business activities by AI agents. Multiple corporate agents continuously carry out business activities, including B2B transactions, to evaluate decision-making and economic behavior in environments where demand, supply, prices, inventory, credit transactions, payment delays, and business costs fluctuate. Rather than completing single tasks, it observes behaviors such as long-term profitability, negotiation tactics, pricing strategies, and failures in cooperation or competition, and is positioned as pioneering research on business processes, internal controls, risk management, and the future of auditing in the era of AI agents.
Comprehensive Analysis
The characteristics seen from the topics on June 26, 2026, reveal that the domestic AI agent market is clearly shifting from "general-purpose chat" to "vertical agents embedded in business processes." In addition to high-impact efficiency areas like log analysis and sales support, it is significant that implementation is progressing in areas requiring high reliability, such as healthcare, auditing, and manufacturing. While connections to existing foundations like MCP, Salesforce, Teams, and AWS are expanding, how to design authority management, data residency, explainability, and human-in-the-loop approval is becoming the core of competitiveness.
Future Points of Interest
The value of AI agents is entering a stage where it is evaluated not by individual performance, but by how safely they can connect to existing business foundations such as AWS, Salesforce, Teams, and MCP.
In high-risk areas such as healthcare, auditing, and labor management, the "branch design between AI judgment and human approval" will be a critical issue that determines the feasibility of adoption, rather than full automation.
As seen in the transfer of manufacturing technology and the knowledge-ization of chat history, AI agents are strengthening their role in turning corporate tacit knowledge into assets, not just improving business efficiency.
With the progress of MCP support and analysis database integration, AI agents will evolve from natural language interfaces to an execution layer that operates business systems.
The spread of wearables and personal AI memory is likely to spark new social debates regarding the balance between the convenience of conversation recording and privacy protection.
