Hello followers! Today, we explore an exciting leap in nuclear tech powered by AI. Atomic Canyon is on a mission to bring the speed of AI to nuclear industry document management. Imagine searching through billions of pages with just a few clicks! That’s what this startup is all about.

Founded a year and a half ago by healthcare veteran Lauderdale, Atomic Canyon is tackling the nuclear industry’s document overload. In places like Diablo Canyon, there are about 2 billion pages of technical documents. Local workers who are like coaches for the plant’s team inspired Lauderdale to find a smarter way to manage all that paper.

With initial funding from Lauderdale himself, the startup developed AI tools to help engineers and compliance staff quickly find necessary documents. Their breakthrough came when they secured a deal with Diablo Canyon late last year, sparking interest from other nuclear operators and pushing the company to raise a $7 million seed round.

This funding was led by Energy Impact Partners, with other investors like Commonweal Ventures, Plug and Play Ventures, and Wischoff Ventures jumping in. The goal? To refine their AI models so nuclear professionals can navigate mountains of data swiftly and accurately.

In testing, Atomic Canyon faced challenges: AI would sometimes hallucinate or produce inaccurate results with nuclear terms. To fix this, they partnered with Oak Ridge National Laboratory, gaining access to massive GPU compute hours on the world’s second-fastest supercomputer. Now, their AI uses sentence embedding techniques to search and index documents effectively with retrieval-augmented generation (RAG), making responses more trustworthy.

Right now, they’re focusing on search, which they see as the foundation. They plan to extend their AI’s capabilities to create draft documents with proper references, always keeping humans in the loop for safety and accuracy. The immense number of nuclear documents means their journey has only just begun.