Genelith Technologies Inc., a Tokyo based developer of custom artificial intelligence solutions for manufacturers, has been selected for the second term of the Tokyo Metropolitan Government's PoC Ground Tokyo program in fiscal 2026. The company, formerly known as CyberX JP Inc., will conduct a proof of concept for an AI system that automates design drawing checks of three dimensional CAD data. The initiative aims to reduce rework and preserve expert knowledge while also inviting manufacturers in Tokyo to join the validation.
Addressing the Final Quality Gate in Design
In manufacturing design, a drawing check is the last quality assurance step where experienced designers visually confirm that a 3D model meets all requirements. The retirement of skilled engineers makes this knowledge harder to transfer to younger teams. When oversights are detected only in downstream processes such as mold production, they lead to costly rework and delays.
How the Automated Design Review AI Works
The proof of concept uses a vision language model to understand 3D models, design requirements and cross sectional views together. It outputs a compliance judgment and a report with supporting reasons. A geometric calculation engine derives exact dimensions from CAD data, while the AI focuses on what to inspect and what to question. Final decisions remain with human designers since the AI provides supporting information.
Invitation to Tokyo Manufacturers
Genelith Technologies is recruiting multiple Tokyo based manufacturers to provide real design data and evaluation support. Suitable candidates use 3D CAD tools such as CATIA, NX, SolidWorks or Creo and maintain internal design standards or checklists. Participants should have past reviewed projects available and be able to assign one or two designers for about one and a half months. The company prioritizes automotive parts, mold and machining businesses, especially those with high mix low volume and reuse oriented design.
Partner Benefits and Data Protection
Participating companies receive a validation report based on their own data, including third party evaluation of accuracy. They also receive records of design requirements, AI judgments and supporting rationale as an initial data asset for skill transfer. There is no cost to participants within the scope of Tokyo metropolitan support. Genelith will sign non disclosure agreements, can deploy in cloud or on premise environments, and will not use input data for AI training.
Application Period and Contact
The application period runs from September 25 to October 25, 2026, with October 25 as the first deadline. Interested companies can apply through the contact email sales@genelith.co.jp or the inquiry form on the company website. Genelith welcomes firms that can contribute design data and evaluation feedback for the proof of concept. Early contact is encouraged because the project requires close coordination with each design team.
Executive Commentary
Noriaki Tsunoda, CEO of Genelith Technologies, described drawing checks as the final bastion of quality assurance supported by experienced designers. He noted that many workplaces lack enough time to transfer this judgment to the next generation. He said the project will test whether AI can first review designs, experienced engineers can correct the output, and the resulting records can feed future decisions. The company plans to report both what was achieved and what still needs improvement.
Program and Company Background
PoC Ground Tokyo is a Tokyo Metropolitan Government initiative that supports startups in concept verification by providing funding for validation costs and matching them with partner organizations. Ten projects were selected for the second term of fiscal 2026. Genelith Technologies was founded on April 25, 2024 and is headquartered in Roppongi Hills Mori Tower in Minato, Tokyo. Its business focuses on custom AI development for manufacturing including design support, visual inspection and defect response.
The selection of Genelith Technologies highlights a practical effort to apply AI where expert knowledge is becoming scarce in manufacturing. By combining visual language models with exact geometric calculation, the project aims to support rather than replace human designers. If successful, the proof of concept could give Tokyo manufacturers a repeatable method to reduce rework and preserve critical design expertise.