I just got back from NIRMA’s 50th Anniversary Symposium in San Antonio, where I had the chance to give two talks: “Don’t Let AI Break Search” and “Maximizing the Value of Your Electronic Content.” Fifty years is a real milestone for an industry association, and spending three days with the people who manage nuclear recordkeeping and information governance gave me a clearer picture of where this industry is headed and what it needs from its technology partners to get there. A few things stuck with me, and most of them apply well beyond the nuclear industry. 

An industry that takes its work seriously

Nuclear energy is one of the more genuinely exciting industries to spend time in right now. The people in that room have real pride in their work and a real appetite for new technology, whether that’s excitement for the small modular reactors coming online or AI tools finding their way into engineering and compliance workflows. That combination of rigor and curiosity isn’t unique to nuclear. We see the same pattern with other organizations that take the time to thoughtfully evaluate new technology rather than rush at it: keep the bar for accuracy high, but stay genuinely willing to adopt new tools once they’ve earned that trust. 

Demand is rising, and trust is the price of admission

Energy demand is climbing, driven in large part by data centers, AI infrastructure, and volatility in other energy markets, and nuclear is positioned to be a major part of meeting that demand. But “nuclear” is still a loaded word for many, and being trusted with that role isn’t automatic. It has to be earned through a laser focus on safety and transparent communication with regulators, government bodies, and the public: being able to show where the evidence is for a given claim or action, who approved a change and under what authority, and whether the organization can defend that decision years or decades later. 

Financial services, healthcare, and other regulated industries answer a version of the same question every time an examiner or auditor walks in. The details differ, but the underlying demand is the same – prove it now and be able to prove it again years or decades later. What changes as AI enters the picture is how that proof gets built and maintained. Regulatory and records management teams need a seat at the table when AI is implemented, because they’re the ones who will eventually have to re-substantiate decisions and claims that were based on AI-assisted search and discovery, long after the systems that produced them have moved on. 

Cautious by design, curious about what’s next

Nuclear is, appropriately, one of the most conservative industries when it comes to adopting new technology. Much of its information is export-controlled and subject to 10 CFR Part 810, so the move to cloud, for example, has been measured and deliberate, and rightly so. Don’t mistake deliberate for uninterested, though. I heard real, active interest in how AI could work within that reality: building AI solutions on nuclear-specific corpora and datasets, supplemented by internal content, and working with partners who understand the compliance environment and can support traceability, transparency, and regulatory defensibility. 

That’s a blueprint every industry evaluating AI can learn from: embrace the technology, but vet it carefully, question its outputs, plan appropriately, and understand the full implications before scaling it. It requires more time up front, but a controlled, deliberate approach pays for itself many times over in project success. 

Where this connects to what we do

That same principle, doing the upfront work before you scale, is exactly where this connects to what we do. Getting content ready for AI has less to do with which AI tool you pick and more to do with the state of the content underneath it. If records are disorganized, duplicated, or poorly related to one another, AI doesn’t correct that, it just returns the same mess faster and with more confidence attached to it. The real work happens upstream: getting content organized, properly classified, and governed before anything gets connected to a search or AI layer. That’s what a modern content platform gives an organization that a network drive or legacy repository never could: the metadata, structure, and security controls to keep content current and trustworthy. Whatever gets connected to it – search today, AI tomorrow – has something reliable to work with instead of noise. That’s the gap Veladocs is built to close: giving organizations a modern layer to organize and govern their content so it’s actually ready for what comes next. We hear the same requirement from manufacturing, pharma, and financial services customers who want AI-readiness without giving up control over their content. 

None of this is unique to nuclear. It’s the same shape of problem we hear from records managers across regulated industries: decades of documentation, real consequences for getting retrieval wrong, and a growing expectation that all of it should be as accessible to AI as it is to a person. Nuclear just makes the stakes very clear.

Fifty years in, NIRMA’s community has spent decades figuring out how to make records defensible over the very long term. If your organization is asking the same question, whether you’re in nuclear or anywhere else with long-lived, high-stakes documentation, we’d welcome the conversation.