Domestic AI Agent Trends (August 11, 2026 Issue)
Update Date: 2026/8/11
Executive Summary
The domestic AI agent trends as of August 10, 2026, have expanded from the introduction of individual tools to organizational design, business system integration, and wide-area deployment in the public sector. NEC has established a new four-tier autonomous department where AI takes on major organizational roles, and ExaWizards announced that its service for local governments has expanded to approximately 60% of prefectural governments nationwide. In the hospitality sector, reservation processing via MCP integration is advancing; in construction, estimate and budget editing involving human approval is being implemented; and within companies, company-wide agent utilization starting from Slack is progressing. In the financial sector, Chiba Bank has presented a large-scale future vision, while Yamagata Bank has announced that it will first roll out RAG search company-wide, indicating a need to evaluate current implementations and future goals separately.
*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 NEC establishes an autonomous organization ranging from AI department heads to AI employees
📎 Source: NEC Press Release
On August 1, NEC established the "Corporate AI & Workforce Department," where AI takes on roles ranging from department heads to employees. It consists of four tiers: AI Department Head, AI Board, AI Manager, and AI Employee, with AI employees generated and appointed to perform tasks according to business needs. Humans control final evaluation, decision-making, and quality/governance. In internal trials in July, the system consistently handled management analysis, simulations, and risk sign detection, reducing the required time to approximately one-seventh. It is attracting attention as an "AI organization" model that integrates distributed agents.
2️⃣ ExaBase AI for Local Governments expands introduction to approximately 60% of prefectural governments nationwide
📎 Source: ExaWizards Inc. Press Release
ExaWizards and Exa Enterprise AI announced that "ExaBase AI for Local Governments," which supports LGWAN, has been introduced and is being used in approximately 60% of prefectural governments nationwide. In six areas across the country, including Gunma and Kanagawa prefectures, joint procurement bundling municipalities is also progressing. In addition to AI agents that autonomously execute slide material creation, minutes creation, and legal research, it is equipped with internal document-linked RAG and a time-reduction dashboard. About 80% of the adopting local governments are using it on a full-scale basis, and education/retention support and wide-area procurement are driving its adoption, not just its features.
3️⃣ Acti-v-ryu begins offering "talkappi ChatAgent," which handles even accommodation reservations
📎 Source: Acti-v-ryu Inc. Press Release
Acti-v-ryu began offering "talkappi ChatAgent," an AI customer service agent for accommodation facilities, on August 10. It integrates a multilingual chatbot and an autonomously updating knowledge base, connecting to PMS, reservation engines, and site controllers via MCP. In addition to answering facility inquiries, it is designed to continuously process reservation confirmations, vacancy inquiries, and accommodation bookings within the chat. Through 24/7/365 response, it aims to reduce phone and email correspondence and expand sales opportunities. However, since some response scopes such as reservation changes and online payments include features under development, the current scope of provision must be confirmed individually.
4️⃣ Concrew makes estimate and budget management AI agent mobile-compatible
📎 Source: Concrew Inc. Press Release
Concrew has expanded the estimate creation and budget management AI agent of "Concrew AI" for small and medium-sized construction companies to a mobile app. When a user gives instructions such as "change the unit price" or "add an item," the AI understands the current estimate and budget and generates proposals for adding, changing, moving, or deleting line items. Changes are displayed as unconfirmed previews, which humans can review and confirm in bulk, or cancel line-by-line or in bulk. Data updated on-site is connected to subsequent tasks such as project management, ordering, scheduling, and cost management. It is a practical Human-in-the-loop type where humans approve before reflecting data in important business records, rather than being fully autonomous.
5️⃣ Givery introduces personal AI agent "Slackbot" company-wide
📎 Source: Salesforce Official News
Givery has introduced the personal AI agent "Slackbot" for business use company-wide, operating Slack as an agentic Work OS where humans and multiple agents collaborate. In sales, it is used for pre-meeting research and automatic updating/analysis of Salesforce using transcripts after meetings, and in marketing, the plan is to execute ad delivery and creative production linked to in-house AI tools via Slack. It uses existing Slack permissions for its security model, lowering the hurdle for company-wide deployment including non-technical staff. On the other hand, revenue improvement and cost reduction are currently expectations, and quantitative introduction effects are subjects for future verification.
6️⃣ Chiba Bank builds sales support infrastructure with an eye on 50 AI agents
📎 Source: Salesforce Japan Co., Ltd. Press Release
Chiba Bank has adopted Salesforce's Agentforce Financial Services, Data 360, and MuleSoft for its new sales support system where humans and AI agents collaborate. In the initial stage, it will centralize customer, project, and activity management, integrate internal data to support agent judgment, and establish API linkages with core systems. The bank has set a goal of operating 50 AI agents and creating 2,000 people's worth of labor during the new medium-term management plan period, but this is not a track record. The new system is scheduled to go live in February 2028, and it is designed so that AI handles routine tasks and escalates to bank employees in situations requiring human judgment.
7️⃣ Yamagata Bank rolls out "neoAI Chat" company-wide; initial use is internal information search
📎 Source: neoAI Inc. Press Release
Yamagata Bank announced that it will begin company-wide use of the corporate AI agent platform "neoAI Chat" from August 2026. Initially, it will be used for information searches referencing internal regulations and business manuals, and in the future, the bank aims to expand business-specific assistants to improve operational efficiency, quality, and knowledge sharing/succession. The platform itself is equipped with dialogue, RAG, OCR, agents, workflows, web search, data analysis, and Office document creation. However, the specific scope of use announced by the bank this time is centered on search tasks, and full-scale operation of business system operations and multi-step execution has not been confirmed.
8️⃣ DGDV invests in "Thomas," an AI founder that autonomously operates businesses
📎 Source: DG Daiwa Ventures Press Release / Y Combinator Official Company Page
DG Daiwa Ventures has invested in the US-based HireThomas, which is developing "Thomas," an AI founder that autonomously launches, operates, and grows businesses. The concept behind Thomas is to continuously execute tasks ranging from problem discovery, software development, and sales to influencer marketing and selecting the next business venture. Rather than building dedicated tool integrations individually, it adopts a "human harness" that provides the same operational interface as a human, including face, voice, PC, smartphone, email, browser, and various applications. While this is not a domestic implementation, it is significant as a move by a Japanese VC to invest in an agent company that conducts autonomous economic activities. Business outcomes and governance will be subject to future verification.
Comprehensive Analysis
From the trends on August 10, 2026, it can be inferred that the competitive axis of the domestic AI agent market has shifted from "model performance" to "designing the safe integration of authorized AI into business operations." talkappi connects to lodging systems via MCP, Concre places human approval before data updates, and Givery and Chiba Bank are integrating business screens, CRM, data, and APIs. NEC has moved beyond individual tools to a stage where departments themselves are composed of AI, with operations and improvements managed centrally. On the other hand, there are differences in maturity. Municipal services have demonstrated the scale of implementation, and NEC has published verification results, but Chiba Bank's figures are future targets, and the initial use cases published by Yamagata Bank are centered on search. Moving forward, it is necessary to distinguish between those that have started provision, are in use, are under construction, or are in the conceptual stage, and to track approval design, authority management, audit logs, and cost-effectiveness.
Future Points of Interest
For NEC's autonomous AI department, the focus is on how much work beyond management analysis can be transferred, and whether the centralized governance model can be replicated at other companies.
For talkappi ChatAgent, attention is on the provision timing for reservation changes, ancillary service bookings, and online payments indicated as future scope expansions, as well as authority management for each lodging system.
For Chiba Bank, it is important to see if they can break down their goal of 50 agents and 2,000 personnel during the new medium-term management plan period, alongside the new system launch in February 2028, into phased KPIs.
In the municipal sector, in addition to LGWAN compatibility and joint procurement, there is a need to develop common evaluation criteria that allow municipalities to compare time savings and operational quality.
In the cases of Concre and Givery, the scope of remaining human approval, methods for inheriting existing permissions, and the design of cancellation and audit procedures in the event of operational errors will determine company-wide deployment.
