In the headquarters of Owens Corning, a global building materials manufacturer, employees gather around poster boards, excitedly sharing their latest AI experiments. It looks like a seventh-grade science fair, but these presenters aren't students. They're engineers, safety managers, and scientists who've discovered ways to transform their daily work with AI.
This scene represents something profound happening in traditional industries. While tech companies dominate AI headlines, manufacturing giants are quietly transforming operations.
Annie Baymiller, Senior Vice President and CIO, has orchestrated this transformation at Owens Corning not through top-down mandates, but by creating an environment where curiosity thrives. Employees are invited to experiment with AI and regularly share their findings.
Baymiller says, “I need the people who live [this work] every day to say, ‘This is wasteful. Wouldn't it be interesting if I could do it a different way?’ Or, ‘Man, I wish I had the data at my fingertips for my decision-making process.’ We need 25,000 great ideas every day for how processes can be reimagined at the doer level.”
The groundswell beats the mandate
Most enterprise transformations follow a familiar script: Executives announce an initiative, consultants design processes, and employees reluctantly adapt. Owens Corning flipped this model entirely.
"I jokingly say I love that there are a million things happening that I don't have to be pushing," Baymiller says. "If all the great ideas are going to come from me, we're going to be short on great ideas."
This philosophy manifested in what Owens Corning calls "AI and Action" sessions — those science fair-style gatherings where employees showcase their experiments. Some AI experiments end up saving employees minutes, others save hours, and a subset simply make work more engaging.
“We’re tinkering, we’re learning,” Baymiller says. “There’s just a natural kind of curiosity in the organization for people who love stuff like this. And there are probably a thousand use cases that I'll probably never know about where someone took 15 minutes out of their day [to invent a new use for AI], and that’s great.”
The cumulative effect? A workforce that sees AI as an opportunity rather than a threat.
From dusty notebooks to digital insights
One of the most compelling transformations emerged from Owens Corning's R&D department. After 85 years, the company had accumulated mountains of lab notebooks, test results, and experimental data — much of it gathering dust in physical archives. "Every couple of years, someone will ask a product-development question we've probably answered years ago," Baymiller notes.
A senior engineer posed a simple question: “What if we could interrogate decades of research instantly?”
The team digitized years of handwritten lab notebooks and test results, then applied Box AI to create a searchable knowledge base. Now, when engineers wonder whether a particular material substitution has been tested before, they don't have to spend weeks combing through archives. They simply ask AI.
Quickly surfacing those answers prevents costly, redundant research and accelerates innovation.
Making safety predictive, not reactive
Perhaps no application better illustrates AI's general potential in manufacturing than Owens Corning's approach to workplace safety. In an industry where physical safety is paramount, the company has moved beyond reactive incident reports to predictive risk management.
By analyzing years of safety data alongside environmental factors (time of year, shift patterns, production volumes) AI now identifies conditions that historically correlate with increased injury risk. When these patterns emerge, plant leadership receives early warnings to implement preventive measures.
"We want every single person to come to work and leave in the same or a better position as when they came. We take safety incredibly seriously.," Baymiller emphasizes. This isn't just corporate rhetoric. It's a data-driven commitment, powered by predictive AI that can spot danger before it materializes.
The art of dropping breadcrumbs
When it comes to inspiring AI adoption, Baymiller explains,”For me, the best way to lead change is to drop breadcrumbs along the way. What we've been trying to do for the last 18 to 24 months with AI is just keep it front of mind, keep it relevant, share little stories that are examples of someone winning and then feeling the benefit of it.
This “breadcrumb strategy” works because it makes change feel achievable. When a safety manager sees a colleague save two hours weekly with a simple AI tool, the technology suddenly feels less threatening and more like an opportunity.
At any enterprise organization, there will be a mix of people excited about adopting new technology and those afraid of what it might mean. By making AI accessible and celebrating small victories, Owens Corning demystifies the technology for employees.
Building guardrails without building walls
It’s important to note that while encouraging AI experimentation, Owens Corning also prioritizes data governance. The company has established two key bodies:
- An AI Community of Practice for innovation
- A Responsible AI Group for oversight
The Community of Practice operates on a hub-and-spoke model, with IT providing technical expertise while business units drive use cases. The Responsible AI Group, comprising IT security, data architecture, and legal representatives, ensures experiments proceed safely and ethically.
"Set your own guardrails around technology that allows innovation to happen at scale for everyone," Baymiller advises. This balance between freedom and control enables rapid experimentation without compromising security or compliance.
3 lessons for traditional industry leaders
Based on Owens Corning's journey, Baymiller offers three critical pieces of advice for CIOs in traditional industries:
First, fix your data foundation. "If you haven't already run fast and hard at your data, especially your structured data, get that in a position where it's accurate, it's trusted, and it can be used," she emphasizes. Without trustworthy data, AI capabilities built on top will never gain user confidence.
Second, establish clear boundaries. Rather than relying solely on policies, build technical guardrails that prevent accidental misuse while enabling innovation. This approach provides peace of mind for both IT leaders and end users.
'Third, become a storyteller. "Don't wait for perfection two years from now," Baymiller advises. "Celebrate the little milestones, celebrate the ideas that turned into proof of concepts, even if they still get killed because they don't add value."
Building on a foundation of AI experimentation
Owens Corning is now shifting from experimentation to transformation. The groundswell of employee-led innovations is now being channeled into systematic process redesign.
This isn't disruption for its own sake, but a recognition that incremental change won't unlock AI's full potential. Owens Corning's story challenges assumptions about AI adoption in traditional industries and proves that real AI progress doesn't require Silicon Valley DNA or massive technology budgets. Instead, it calls for something more fundamental: leaders willing to trust their people, employees willing to experiment, and organizations willing to learn from both successes and failures.
As manufacturers, utilities, and other traditional industries grapple with AI adoption, they'd do well to study the science-fair approach Owens Corning has instituted. There, among the poster boards and enthusiastic demonstrations, the future of industrial AI is being written one breadcrumb at a time.
Read more about how Owens Corning uses Box for Intelligent Content Management.
