Tektome joins Japan MLIT AI building review programme

Tektome Inc. has joined a Japanese government-backed programme to test AI in building approval reviews and set common benchmarks for judging tool reliability. The project could shape how Japan evaluates AI for building confirmation ahead of planned BIM-based reviews from 2029.

Why it matters: - Japan is trying to make building confirmation reviews faster, more consistent and more transparent. - The programme aims to create shared benchmarks so private-sector review bodies can compare AI tools on the same basis. - The work could reduce human error, ease the burden on reviewers and applicants, and influence how AI is procured for building administration.

What happened: - Tektome Inc. said it is participating in the Ministry of Land, Infrastructure, Transport and Tourism’s FY2026 Comprehensive Programme for Promoting DX in Building Administration. - MLIT notified Tektome affiliate Incubit, Inc. on 23 April 2026 that Incubit’s proposal had been selected. - The selected project is titled “Theme 4: Study on Promoting DX in Building Administration.” - The programme runs under Japan’s FY2026 fiscal year, which spans April 2026 through March 2027.

The details: - The MLIT programme supports efforts to unify building administration procedures and improve the efficiency and quality of reviews and inspections. - Building administration covers the statutory review of building plans for compliance and the inspection of buildings under Japan’s building regulations. - The project will study how AI can be used in the building confirmation application process and how advanced those tools are today. - The project will also build benchmarks and an evaluation framework to compare the reliability of different AI solutions on a common basis. - Building confirmation is the statutory review, before construction begins, of whether building plans comply with Japan’s Building Standards Act. - The project will trial AI tools for tasks including checking documents against formal requirements, identifying the documents and sections that need review, and verifying consistency between drawings and application forms. - The work will also survey benchmarks for general-purpose generative AI and examples of AI performance evaluation in Japan and overseas. - The project will define evaluation criteria including accuracy and explainability, then create benchmark metrics and test questions for AI tools handling building confirmation applications. - Tektome’s role is to identify needs and challenges in building confirmation application and review work through interviews with designated confirmation and inspection bodies. - Those designated bodies are private-sector organisations authorized by national or prefectural governments to carry out building confirmation and inspections, and they handle most building confirmations in Japan. - Tektome will also identify review items, work processes and AI needs in current 2D-drawing-based review work. - Tektome will prepare for BIM data review, which MLIT aims to introduce for building confirmation from spring 2029. - During the trials, Tektome will assess how applicable and useful the tools are in real review work and use the feedback to improve the evaluation framework and test questions.

Between the lines: - The project is not just testing AI performance; it is trying to define what trustworthy AI should look like in a regulated public workflow. - The emphasis on explainability suggests regulators want tools that can justify results, not only generate them. - The benchmark approach also points to a procurement problem: public and private reviewers need a common way to compare vendors without relying on marketing claims. - The move toward BIM-based review shows Japan is already planning for a more digital approval process, and this project appears designed to bridge today’s 2D workflow with that future system.

What's next: - The project will run interviews and other fact-finding to identify use cases and operational needs. - It will then carry out trials of AI tools intended to make review work more efficient. - The project will compile a draft proposal, trial results, benchmark questions and benchmark scores. - Those outputs are meant to give private-sector review bodies and businesses a fair and transparent way to assess AI performance and reliability. - Tektome said its work will contribute to reducing reviewer and applicant burden and improving the quality and transparency of building administration. - Tektome is an AI solutions company founded in 2023 and based in Tokyo. - The company provides the Tektome platform for architectural design using company-owned design data to support knowledge transfer and productivity. - The company’s website is Tektome.