Generative AI is no longer a future concept reserved for tech companies. It is quickly becoming part of the operational toolkit for manufacturers across South Carolina – and leaders are working to separate meaningful opportunity from marketing hype.
On February 3, OpExChange hosted a virtual session titled Demystifying GenAI, drawing one of the strongest webinar responses in the network’s history. The high turnout reflected a growing reality: manufacturers are curious about generative AI, but they are also cautious. They want clarity, not buzzwords.
The session was led by Cameron Duncan, Co-Founder of Hallian Technologies. With nearly 15 years of manufacturing experience prior to founding his company — including roles at Proterra, Gestamp (Union, SC), and GE Power — Cameron brought an operational lens to a topic often dominated by technical jargon.
His message was clear: generative AI is powerful, but it is not magic. And in manufacturing, context matters.
What Generative AI Actually Is
Cameron began by stripping away the mystique.
At its core, generative AI is a prediction engine. Powered by large language models (LLMs) such as ChatGPT, Copilot, and Gemini, it generates text, images, and code by recognizing patterns in vast datasets and predicting what comes next.
To make the concept tangible for manufacturers, Cameron drew a comparison to plant-floor vision systems. Just as a vision system is trained on thousands of “good” and “bad” parts to predict pass or fail outcomes, generative AI predicts likely language responses based on patterns in data.
It does not reason the way humans do. It predicts.
Understanding that distinction helps explain both its strengths and its risks — including hallucinations, where the system produces confident-sounding answers that may be inaccurate if context or data is incomplete.
For manufacturers accustomed to disciplined problem-solving, that clarity was critical.
Where GenAI Shows Immediate Promise
Rather than focusing on futuristic automation, Cameron emphasized practical, near-term applications – especially in knowledge-heavy areas where teams spend significant time searching, summarizing, and documenting information.
Several use cases resonated strongly with attendees:
Maintenance and Reliability
- AI-powered assistants referencing manuals, SOPs, work orders, and equipment history
- Faster troubleshooting and reduced unplanned downtime
- Auto-generated inspection checklists
Engineering and Technical Teams
- Embedded access to standards, drawings, and legacy documentation
- Faster development of work instructions and DFMEAs
- Democratizing expert knowledge across teams
Quoting and RFP Responses
- Analyzing historical proposals and customer requirements
- Drafting 70–80% complete responses in minutes rather than days
- Allowing teams to focus on judgment instead of boilerplate writing
Throughout the discussion, Cameron reinforced a key point: generative AI does not replace people — it amplifies them. Its greatest value lies in removing low-value friction and freeing skilled professionals to focus on higher-level thinking.
The Unavoidable Foundation: Data Quality
One of the most important – and least glamorous – messages of the session centered on data.
Successful generative AI deployments are less about flashy tools and more about disciplined preparation. Cameron noted that the majority of implementation efforts typically involve organizing, curating, and maintaining high-quality data.
Manufacturers often hold their most valuable information in unstructured formats:
- PDFs
- PowerPoint presentations
- Excel files
- Quality manuals
- Work instructions
The encouraging news is that generative AI excels at working with unstructured data — if it is well organized and governed.
Cameron encouraged leaders to view their data as a strategic asset, much like critical production equipment. It requires ownership, maintenance, and continuous improvement.
“Poor data in,” he emphasized, “will simply produce poor results faster.”
Security, IP, and the Risk of Shadow AI
Security was one of the most discussed topics of the session — and for good reason.
As generative AI tools become widely accessible, employees may upload sensitive information into public platforms to solve immediate problems. While often well-intentioned, this “shadow AI” behavior can create serious risks related to intellectual property and confidentiality.
Cameron outlined key considerations:
- Public AI tools may retain and train on uploaded data
- Proprietary processes can unintentionally leak
- Liability protections from public platforms are limited
He encouraged leaders to ask direct questions about data handling, retention policies, and deployment models. Options range from off-the-shelf public tools to fully private, closed systems tailored for enterprise environments.
The takeaway was balanced but firm: innovation must be paired with intentional governance.
Adoption: A Leadership Responsibility
Technology alone does not drive value – people do.
Cameron emphasized that successful AI initiatives require executive understanding, clear guardrails, and a culture that allows thoughtful experimentation. Many of the most compelling use cases originate from employees closest to the work — when they are empowered to test ideas responsibly.
Several plant leaders shared that the session helped them move from uncertainty to clarity – not because every question was answered, but because they now understood how to approach the conversation more strategically.
That shift in mindset may be as important as any specific tool.
Looking Ahead
Generative AI is not a passing trend. It is rapidly becoming part of how business gets done.
Like earlier waves of automation, lean transformation, and digital integration, early adopters are gaining advantages – reducing administrative burden, accelerating decision-making, and enabling teams to focus on problem-solving rather than paperwork.
At the same time, manufacturers who rush forward without governance risk exposing intellectual property or eroding trust.
The competitive differentiator will not be who experiments the fastest, but who learns the most intentionally.
As Cameron noted during the discussion, within a few years AI will no longer be a debated topic – it will simply be embedded into daily workflows. The organizations that take time now to understand it, structure it, and align it with their operational culture will be better positioned to lead.
The central question remains a familiar one for manufacturers:
How do we use the right tools, in the right way, to make our people more effective?
That is a conversation worth continuing.
About OpExChange
OpExChange, sponsored by the South Carolina Manufacturing Extension Partnership (SCMEP), is a peer-to-peer network of manufacturers and distributors across South Carolina. The organization fosters collaboration through plant tours, benchmarking sessions, leadership forums, and practical knowledge sharing — all within a sales-free environment focused on strengthening the state’s manufacturing competitiveness.
More information about upcoming events and participation can be found at www.OpExChange.com.
Be the first to comment on "Demystifying Generative AI in South Carolina"