R3 Institute Developing 'gusuku Fluxon' to Securely Connect Generative AI with kintone

R3 Institute (Headquarters: Osaka, Japan; CEO: Akira Nishimatsu) announced the ongoing development of 'gusuku Fluxon,' a new service designed to securely utilize kintone data from generative AI and MCP (Model Context Protocol) clients.

gusuku Fluxon aims to provide a 'managed Remote MCP server' between generative AI and kintone, allowing administrators to implement additional control and auditing while respecting the individual user's original kintone access permissions.

The official launch date for gusuku Fluxon is currently undecided. The company plans to exhibit the concept and partial features of the service at 'kintone x AI Expo 2026,' to be held on June 8, 2026.

Background As the use of generative AI in business grows, demand has increased for utilizing operational data stored in kintone via AI.

However, when AI clients connect to business data, challenges arise regarding the handling of authentication credentials, per-user permission management, control over data shared with AI, and auditing of access logs.

gusuku Fluxon aims to address these issues by placing a secure control layer between generative AI and kintone, creating an environment where data utilization can be pursued with greater peace of mind.

Concept gusuku Fluxon is not a service that provides AI itself; rather, it functions as a gateway for generative AI and MCP clients to safely utilize kintone data.

Key pillars: 1. Respecting User Permissions: Access to kintone is based on the user's own OAuth authorization. Authentication credentials are not passed directly to the AI client but are managed securely on the Fluxon side. 2. App-Level Control: The service aims to control operations such as view, add, edit, or delete from generative AI at the kintone app level. 3. Administrative Control: Administrators can apply additional restrictions on apps or operations that they do not want to be accessed via AI, even if the user has permissions within kintone itself. 4. Accountability via Audit Logs: Actions taken—who accessed which kintone app from which AI client—are recorded as audit logs for administrator review.

Key Planned Features - OAuth-based per-user authorization. - Connectivity from Remote MCP-compatible AI clients. - Granular control over apps, including field-level reference control and masking. - Administrator and user web UIs. - Audit log display, search, and CSV export.

Exhibition at 'kintone x AI Expo 2026' R3 Institute will showcase gusuku Fluxon at 'kintone x AI Expo 2026' on June 8, 2026. Official release date and pricing are currently TBD.