HEROZ ASK, an Enterprise Generative AI SaaS, Begins Supporting "MCP" to Enable Integration with Popular Business Tools like Notion, Box, and Slack

HEROZ Corporation (Headquarters: Minato-ku, Tokyo; CEO: Takahiro Hayashi; hereinafter "HEROZ") announced that its enterprise generative AI SaaS "HEROZ ASK" has begun supporting "MCP (Model Context Protocol)," a new standard connection protocol that safely and flexibly connects AI with external services, starting May 1, 2026. This enables integration with major business tools such as Notion, Box, and Slack, creating an environment where AI agents can autonomously utilize external tools to perform tasks.

Through this initiative, HEROZ aims to achieve "AI Cowork," where people and AI work together as colleagues. We will promote the creation of an environment where AI is not just a "tool that answers" but autonomously moves as a team member within the workflow, allowing humans to concentrate on more creative and strategic decision-making.

Background of MCP Support

The use of generative AI in companies is rapidly evolving from supporting isolated tasks like document creation, information retrieval, and summarization to automating complex business processes across multiple systems and tools. However, until now, integrating generative AI with various internal and external business tools often required individual system development and complex configurations, leading to high hurdles for implementation and operation.

Against these challenges, "MCP (Model Context Protocol)" is a new connection standard that has been gaining attention in recent years. MCP serves as a common "bridge" for connecting AI with external tools and services. By supporting MCP, integration with multiple external services can be achieved safely and efficiently through a standardized method. HEROZ has decided to support this feature to further deepen the "practical application of generative AI" in our customers' business environments.

What is AI Cowork?

AI Cowork is a new way of working with AI, where multiple specialized AI agents collaborate in real-time as a team across the entire business operation.

Until now, AI tools primarily followed a single-agent model where "one AI answers one question." However, with this model, in complex tasks spanning multiple tools, humans were still left with the burden of coordination, such as giving instructions to AI, checking results, and integrating them.

AI Cowork changes this paradigm. By having multiple AI agents divide roles and operate in parallel, autonomously coordinating with external tools, humans are freed from the coordination burden of a project manager and can focus on essential tasks such as judgment, creativity, and communication.

HEROZ aims for an "AI Cowork" world where AI evolves from being a "tool to use" to a "partner to work with." This means not only conversing as an AI assistant but also processing multiple tasks in parallel while integrating with external tools, supporting human judgment, and fundamentally transforming organizational productivity—this MCP support is a crucial first step HEROZ is taking toward realizing AI Cowork.

Overview of Main Functions of MCP Support

① Centralized Management of External Services (MCP Admin Screen)

Administrators (group owners) can add and manage integrated external services from the newly established MCP admin screen. They can choose between two types: services commonly used throughout the organization (pre-configured MCPs) and services individually configured for specific groups or purposes (manual MCPs).

Furthermore, in organizations where multiple teams or departments use HEROZ ASK, by implementing a mechanism where subgroups (group units per department/team) apply for integration and the overall organization administrator approves it, secure and flexible operation can be achieved simultaneously.

② Automatic Utilization by AI Agents

In the agent function (AI Agent Builder/Project function) of "HEROZ ASK," activated MCPs can be set as tools. Once set as a tool, the AI will judge the situation during a chat and automatically call external services to process tasks. Even if the person in charge does not know how to operate the tool, the AI will act on their behalf, making business flow automation simpler to achieve.

③ Direct Use from Chat Screen

Even in regular chats (conversational projects) that do not use the agent function, by selecting an activated MCP from the tool menu on the chat screen, answers and processes linked to external services can be performed.

④ User's Own Connection Management

Individual external service connections and disconnections can be easily managed from the user menu under "Settings → Connected Apps."