#Japan #Tokyo #AI Platform #Terrasky #mitoco Buddy

TerraSky's Mitoco Buddy Upgrade: Key Features and Innovations

On July 14, 2026, TerraSky, a prominent firm based in Chuo, Tokyo, will release an upgraded version of its AI platform, mitoco Buddy, which is compatible with the Model Context Protocol (MCP). This latest version aims to improve the platform's utility with a range of new functionalities and enhancements.

One of the standout features in this upgrade is the Auto Router, designed to streamline the process of model selection based on user requests. The Auto Router intelligently analyzes the complexity and cost of a request, and automatically selects the most suitable AI model. This eliminates the need for users to manually choose models, ensuring optimal performance tailored to specific tasks. The Auto Router categorizes requests into three distinct levels:

1. Light (e.g., claude-4.5-haiku): Best for short, standalone requests that can be answered in a single interaction.
2. Balanced (e.g., claude-4.6-sonnet): Suitable for business communication tasks, allowing users to engage without significant modifications to their communication style. This level is also efficient for brief conversations or interpreting results from read-only tools.
3. Heavy (e.g., gpt-5.4): Ideal for complex coding or data analysis tasks, and lengthy discussions that require nuanced context. This mode is crucial for high-stakes tasks that may incur significant losses from errors, as well as for operations that necessitate many tool calls.

Moreover, the model selection logic will undergo continuous improvements, ensuring that user data remains confidential and is not referenced during the selection process.

Enhanced File Management with SharePoint

The updated mitoco Buddy also introduces functionality for incorporating files from SharePoint into knowledge bundles structured for AI comprehension. This upgrade will prevent the manual download and upload of documents, allowing users to synchronize easily with their agents through straightforward actions.

Governance and Quality Assurance

Another significant addition is the Agent Publication Approval Workflow. This feature requires administrator or owner approval before agents created within an organization can be shared broadly. By instituting a third-party verification step prior to publication, the workflow enhances content quality, security, and governance protocols.

Knowledge-Centric Features

An enriching feature allows the AI to suggest skills based on repeated tasks or complex interactions identified within general chat. This simplifies processes by eliminating redundant information entry, enabling users to execute complicated operations with a single click.

Dynamic skill generation is another exciting aspect, where users can create necessary skills spontaneously within the chat environment, minimizing the need for file uploads and enabling seamless execution of tasks within the chat context.

Tailored MCP Tool Integration

The latest version also proposes dynamic settings for MCP tools based on the chat's content and user requests, automatically searching and suggesting appropriate tools for the task at hand.

Improved Chat History Functionality

Users can now search their chat history naturally, asking questions like, “What conversations did I have this week?” The AI will summarize relevant discussions from the past three months, allowing users to re-engage with previous conversations seamlessly.

Secure External Integration

The upgrade also features user-friendly management of OAuth tokens for external applications, ensuring that authentication information remains secure and optimized. New MCP connection URLs for external services will also be issued, simplifying the integration process with internal systems.

Additionally, mitoco Buddy enhances collaboration capabilities with new services like Azure Files and IBM Netezza, allowing users to access company files directly from chat interfaces and process large datasets efficiently.

Finally, mitoco Buddy will now support the latest AI model, Claude Sonnet 5, enabling more sophisticated interactions and natural dialogue capabilities.

For detailed information about these features and enhancements, refer to the release notes available here.

About mitoco Buddy

The mitoco Buddy platform assists businesses in leveraging data and automating operations, compatible with around 50 different services, including Salesforce and Slack. Users can create custom agents that enhance automation, taking full advantage of their existing multi-cloud environments.

Please note that all company names and service names mentioned are registered trademarks of their respective owners.