Articul8 AI and the American Society of Mechanical Engineers (ASME) have announced the first generative artificial intelligence model built specifically around engineering technical standards, with a direct focus on the nuclear and aerospace sectors. The agreement, made public on July 14, 2026, involves training a domain-specific model on the entire ASME normative corpus — codes and standards that govern the design, construction, and operation of nuclear plants, pressure systems, aerospace components, and much more.
ASME continuously publishes and updates one of the broadest and most widely recognized regulatory portfolios in the world, spanning pressure technology, power plants, nuclear components, and piping systems. Its codes are accepted in over 100 countries as the benchmark for government safety regulations. Until now, navigating this body of knowledge required specialized human expertise and hours of consultation. The model developed with Articul8 aims to make this information queryable in real time, with traceable and verifiable responses.
Articul8 is no generic startup. Founded in January 2024 as a spin-off from Intel in partnership with DigitalBridge Group, the company has focused from the outset on regulated industries: energy, manufacturing, aerospace, and semiconductors. Its A8-Energy model, developed alongside EPRI on more than 10,000 specialized energy-sector datasets, has already demonstrated 96.9% accuracy on regulatory and infrastructure tasks, compared to 71.3% for general-purpose models. That gap is far from negligible when working with technical documentation that carries direct safety implications. In January 2026, the company closed a Series B round at a valuation exceeding $500 million.
The partnership with ASME follows a clear rationale. Regulatory codes are living documents, subject to periodic revision, structured around dense and often interdependent technical specifications. General-purpose models tend to fall short precisely here — not due to any lack of language capability, but because of the absence of domain context. Articul8 builds its models directly from original technical data, with an architecture designed to ensure full traceability and auditability of every response. In the nuclear sector, where ASME codes are incorporated by reference into regulations issued by the Nuclear Regulatory Commission, this level of precision is not optional.
For the nuclear industry, the practical implications are significant. SMR reactor designers, certification engineers, and regulatory licensing teams all work daily with ASME codes. Being able to query these documents using a model trained specifically on them — rather than relying on manual searches or general-purpose models prone to hallucinating regulatory details — reduces error margins and verification time. Gartner has already recognized Articul8 as a reference player in domain-specific AI for manufacturing and energy. If the model developed with ASME delivers the accuracy levels already demonstrated across other verticals, nuclear engineering will have a next-generation operational tool for regulatory compliance.
