デジタルツイン総合研究所

株式会社デジタルツイン総合研究所 (Digital Twin Research Institute, abbreviated DTRI) is a Tokyo-based technology startup whose stated mission is to build the foundation for industrial deployment of physical AI. The company develops a "field context" platform that semantically connects video, spatial, time-series and business information without requiring enterprises to move confidential data outside their own environment, so that a variety of AI models and robots can reference and coordinate through it.

Its principal product, SHOKEI, was released on 2026-08-05 and is positioned by the company as an on-premises reference and orchestration layer — described as "RAG for physical AI." SHOKEI ingests and normalizes multimodal plant data (video, images, audio, point clouds, 3D models, drawings, documents, equipment logs, sensor time series, quality data and work records), links it to common object IDs, timestamps, coordinates, processes and version information, and represents equipment, parts, people, tasks, events, states, judgments, constraints and results as entities in an ontology and knowledge graph. Rather than embedding company-specific data into models, it retrieves and assembles the context a given model needs at inference time for a specific task, object, location, time and state, and manages shared context and task state so that multiple models, agents, video platforms and robots operate against the same understanding of the site and its safety constraints.

The company frames this as a redefinition of the digital twin: from a model used for visualization, simulation and human monitoring toward a continuously updated knowledge base that AI and robots reference during inference and execution. Targeted application areas include robot and humanoid work execution, visual inspection and process monitoring, equipment maintenance, and linking design/analysis work to manufacturing outcomes.