By Bryan Sapot, Executive Vice President, Nulogy
The conversation around manufacturing AI has no shortage of bold predictions. But after nearly three decades building software for manufacturers, and actively developing and deploying AI-powered capabilities at Nulogy, I’ve learned to pay closer attention to what’s actually happening inside factories than what’s being forecast from the outside.
The honest picture is nuanced, and more interesting than a lot of the hype I know I’m hearing in the market.
Manufacturing AI Adoption Is Real, But Still Early
AI usage in manufacturing software is growing, with PwC reporting that 86% of manufacturers are accelerating investment in AI and automation.
Most of what companies are experimenting with today falls into the realm of AI-enabled product features, such as smarter recommendations, predictive alerts, and automated reporting. However, fully agentic AI (i.e., systems that can reason, plan, and act autonomously across complex workflows) is still in the early phase.
That said, the experiments are already producing some surprising stories.
One of our customers recently told me that a paint line supervisor at their facility had built an app using an AI tool to more efficiently sequence his production line. This person wasn’t a software engineer or a knowledge worker, but a floor supervisor who works in the factory every day and saw a problem he could solve. He figured it out and built something useful on his own.
That kind of grassroots experimentation is the signal I find most meaningful. When frontline workers are independently finding ways to use AI to make their jobs easier, it tells you that the technology has become accessible enough to cross into everyday problem-solving. The question is whether companies will be able to channel that energy strategically.
What’s Blocking AI Adoption in Manufacturing?
In my view, the two biggest roadblocks right now are knowledge and policy.
On the knowledge side, manufacturers are genuinely uncertain about what’s possible. Our Product team consistently hears the same question from customers: “What can we (or should we) actually do with AI?” There’s a strong indication of curiosity, but without clear guidance, that curiosity often stalls out.
On the policy side, manufacturers have legitimate concerns about protecting their intellectual property. Proprietary information such as production data, material specifications, and process parameters are assets that companies have spent years developing. The hesitation to expose that data to LLMs is understandable, and it’s one of the real friction points slowing adoption across the industry.
The path forward isn’t to dismiss those concerns; it’s to address them with clear internal policies that define what AI tools employees can and can’t use, and to provision those tools through official company channels rather than leaving workers to use their own personal accounts.
Where the Real Opportunity for Manufacturing AI Lies
The AI use cases seeing the most traction today are largely office-based: preparing presentations, drafting customer-facing documents, analyzing operational data. These are meaningful productivity gains, but they’re only scratching the surface of what’s possible.
The real value unlock will happen on the factory floor as emerging technologies may it easier for manufacturers to solve problems in real time. Imagine a machine producing out-of-spec product for an unknown reason. Right now, a team has to pull together maintenance histories, material specs, process logs, and tribal knowledge – often under significant time pressure. An AI agent that can synthesize all of that context instantly and propose solutions won’t just save time. It will keep production moving and get answers into the right hands faster, all of which benefits the business.
We’re also innovating with AI at Nulogy. We recently introduced Nora, an AI assistant embedded directly inside Smart Factory and Maintenance Management. Nora lets plant teams ask questions about production performance, downtime, and maintenance orders in plain English – pulling answers from their live data and helping them act on what they find, without bouncing between systems.
Any manufacturer operating across multiple disconnected systems (where employees routinely have to jump between three or four platforms just to complete a single task) stands to benefit enormously from agentic AI. The same is true for companies running legacy ERP systems that are difficult to navigate, customize, and maintain. AI can act as an intelligent layer that simplifies those interactions and reduces the cognitive load on the people doing the work.
Nulogy’s Approach to AI
At Nulogy, our view is that AI in manufacturing has to be grounded in operational reality. That means building capabilities that integrate with how manufacturers actually work: connecting to the systems they already have, respecting the data security requirements they need to maintain, and solving the problems their teams face day in and day out.
It’s also why we created Nora, which directly connects the people on the floor to the data behind them – turning the questions supervisors and operators ask every day into instant, actionable answers.
The companies that will get the most out of AI are those that move past the question of whether to adopt it, and into the harder, more important work of figuring out how. That means investing in education, establishing clear governance, and giving employees structured access to tools that can genuinely help them do their jobs better.
The paint line supervisor building his own sequencing app is a good sign. The goal now is to make sure that kind of initiative has the right foundation behind it.
To learn more about how Nulogy is implementing AI to help manufacturers reduce costs, optimize labor, and improve work, contact our team or book a demo today.
