“OFFICE AGENTIC AI” that drives behavioral change: a new office model where people and AI work together
KOKUYO Co., Ltd. (Head office: Osaka; President and Representative Executive Officer: Hidekuni Kuroda) and TIGEREYE Inc. (Head office: Chuo-ku, Tokyo; CEO: Gaku Kamimura) have filed a joint patent application for “OFFICE AGENTIC AI,” a technology aimed at driving behavioral change and raising intellectual productivity in office spaces.
This technology is an “agentic” AI platform in which AI learns from individual behavioral data and evolves the relationship between people and space through dialogue, evaluation, and optimization. “OFFICE AGENTIC AI” is the result of joint research that combines the multimodal AI framework developed by TIGEREYE with KOKUYO’s office philosophy for an “Autonomous Collaborative Society.”
At the core of this technology, interactions among AIs and between AIs and people are managed with a structured protocol (MCP), and conversation content, intent, and contextual information are vectorized and learned and updated as “affinity vectors.”
This enables AI to autonomously decide “how,” “with whom,” and “for what purpose” it should collaborate, and to provide optimal support tailored to the office environment and each person’s way of working.
In addition, the technology adopts TIGEREYE’s proposed three-layer architecture of a Talk layer, Judge layer, and Match layer:
- Talk layer: generates dialogue with users and AI
- Judge layer: scores and evaluates dialogue content
- Match layer: selects the optimal agents and people
Through this mechanism, AI in the office learns autonomously and optimizes spaces and people’s behavior in real time.
This architecture goes beyond the scope of a conventional AI assistant and is the technical foundation for turning the office itself into “something that thinks and makes proposals.” By analyzing employees’ behavior patterns, meeting data, and communication histories after anonymizing them, the system will be able to propose optimal seating arrangements, meeting times, team compositions, and more.
Key technical points
- Exchange of multidimensional meta-information via MCP (Model Context Protocol)
- Multidimensional evaluation of AI responses using scoring vectors
- Optimal matching of agents and people using affinity vectors
- Dynamic learning of history, results, and behavior using a graph database (GraphDB)
- An autonomous control architecture with three layers (Talk / Judge / Match)
Comment from KOKUYO Co., Ltd.
“We want to evolve workspaces from a mere ‘place’ into an ‘agent that changes behavior.‘
Our collaboration with TIGEREYE is a first step toward shaping the workplace of the future, where the office learns on its own through AI and draws out people’s creativity.”
Comment from TIGEREYE Inc.
“‘Agentic AI’ that does not simply respond but autonomously thinks, evaluates, and guides action is the next generation of intelligent infrastructure. By combining KOKUYO’s spatial design philosophy with our AI control technology, we aim to turn the office into an ‘evolving intellectual ecosystem.’”
Future developments
Based on this technology, the two companies will proceed with implementation as “OFFICE OS” and “OFFICE AGENT” and plan to begin proof-of-concept trials in operations such as meeting room reservations, attendance, information sharing, and well-being measurement. Ultimately, they will realize a platform in which AI understands each employee’s behavior, emotions, and goals and evolves the way people work itself.
KOKUYO × TIGEREYE joint application: OFFICE AGENTIC AI (learn more)
Start by telling us about the challenges you want to solve with AI
Toward workplaces where AI is simply part of everyday work.