When Trey Lauderdale moved to San Luis Obispo, California three years ago, he found himself living in the shadow of Diablo Canyon, the state’s last operational nuclear power plant. As conversations with local plant employees became routine, the Industrial Engineer from the University of Florida began to see striking parallels between his past and a potential future.
Fifteen years earlier, around 2008, Trey had founded Volt, a company with what many considered a “crazy” idea: putting phones in the hands of doctors and nurses for critical hospital communication. At the time, Blackberry ruled the enterprise world, and the notion of using consumer smartphones for accessing electronic medical records, video calls, and emergency alerts seemed far-fetched.
But Trey understood something about risk-averse industries. Hospitals, like nuclear plants, operate under a “do no harm” philosophy. Both require extreme security measures whether protecting patient data under HIPAA or ensuring cyber security in nuclear facilities. Both prefer on-premise solutions over cloud-based systems. And both demand that new technology prove itself incrementally.
In 2019, when Volt was sold, it had become the standard for smartphone-based clinical communication across leading health systems nationwide.
It was then Diablo Canyon that revealed a new opportunity. While nuclear power represented a “modern-day miracle” for clean energy, Trey discovered it was buried under “tremendous red tape and bureaucracy.” The industry generated massive amounts of documentation for regulatory purposes, yet plants struggled to find the information they needed when they needed it.
Atomic Canyon’s approach mirrored Trey’s healthcare playbook that was to begin with non-mission critical applications to build trust. They were not suggesting that AI should operate nuclear reactors, that would have been “insane” and “the plot of Terminator,” as Trey puts it. Instead, they were focused on helping the industry access the data it already has.
Their foundational platform, Neutron, tackles what seems like a straightforward problem but represents a significant industry pain point. The company downloaded approximately 52 million pages of public documents from the NRC’s ADAMS database and used AI technologies computer vision and speech recognition to understand the content and meaning within these documents.
This resulted into an intuitive search interface that contrasted sharply with ADAMS’ Boolean-based search system.
It’s free for anyone in the nuclear industry, though sign-up is required to prevent access by foreign actors or those with malicious intent.
Atomic Canyon secured access to Frontier, one of the world’s fastest supercomputers at Oak Ridge National Laboratory. They have been using this computational power to train “sentence embedding models” teaching AI to understand nuclear language and context rather than simply generating responses.
The company platform Neutron’s search capabilities are impressive. This platform is powered by nuclear-specialized AI embedding models known as FERMI. It achieves unprecedented accuracy in nuclear-specific searches, delivering results that are 40% more accurate and twice as efficient as existing solutions.
Neutron platform is currently being deployed at PG&E’s Diablo Canyon Power Plant, which reduces document search time from hours to seconds across billions of pages of technical documentation.
These models teach AI to understand nuclear-specific terminology, abbreviations, and industry dialects, recognizing that the same equipment might have different names across reactor vendors and that terms carry different meanings in different contexts. This addresses a fundamental problem: generic AI hasn’t “seen enough nuclear words to understand what they mean.”
Rather than building generative AI that might “hallucinate” incorrect information in a safety-critical industry, Atomic Canyon focuses on intelligent search and document understanding.
The company just announced a $7M seed funding round, from Energy Impact Partners‘ Elevate Future Fund, Commonweal Ventures, Plug and Play Tech Center, Wischoff Ventures, Tower Research Capital, and other angel investors.
As reported by TechCrunch, Atomic Canyon’s AI is designed to produce initial drafts of documents, including citations. Lauderdale emphasized that human oversight will always be necessary in the process.