June 29, 2026 . By Kirstyn Brown. Manufacturers and industry experts talked openly about AI adoption: what's working, what isn't and why data and training matter more than the technology itself.
Artificial intelligence took centre stage at FABTECH Canada 2026, as speakers across keynote and panel sessions focused on its growing role in shaping manufacturing operations, while also acknowledging the challenges of applying it on the shop floor.
Held June 9–11 in Toronto, the event brought together fabricators, welders and manufacturing professionals to explore new technologies, connect with more than 300 exhibitors and attend expert-led panel discussions.
Among those discussions, a Day 1 panel on AI-driven automation in custom metal forming and fabrication offered a snapshot of how manufacturers are approaching the technology today. Helen Papachronis, director of project development at NGen, said the industry has spent years building the foundation through sensors, data collection and connectivity.
“About 10, 15 years ago, we really started talking about big data and internet of things and digitization of operations and processes and equipment,” she said. “And AI is just the next step. Now it’s, what do we do with all that digital information?”
The panel also explained how that data is now being used. Papachronis said AI allows manufacturers to move from reacting to problems after they occur to responding as they develop.
“AI is that layer above the digitization, the data collection to start to really automate and see real time operations,” she said. “It’s going to take things from a reactive operation to a proactive operation.”
When asked what he sees as the biggest AI-driven shift on the custom fabrication shop floor today, Sony Pauly of Siemens Digital Industries Software described a convergence of speed, accessibility and intelligence at the machine level.
“There are two pieces that are coming together — the proactive nature and the speed of response,” he said. “The other is generative AI, where anybody on the shop floor is able to interact, command and ask questions and get to a level of detail that wasn’t possible in the past.”
For Greta Cutulenco, CEO and co-founder of Acerta Analytics Solutions, the shift is happening at the system level, as AI expands beyond individual machines to influence entire production processes.
She added that this broader application of AI is enabling more data-driven decision-making across every stage of manufacturing and gradually pushing operations toward greater autonomy, though engineers remain critical to day-to-day operations.
Use cases growing, but still uneven
The Day 1 panel highlighted a range of emerging applications, particularly in areas where manufacturers already generate large amounts of data. Quality inspection, predictive testing and process optimization were among the most common examples.
Cutulenco said companies often turn to AI when dealing with persistent performance issues such as production lines where first-time-through rates are inconsistent, teams are firefighting daily and the root cause isn’t obvious.
She also mentioned testing as another high-impact area. In some cases, manufacturers are spending hours testing individual batteries or fuel cells coming off a line, a significant drain that predictive AI applications are beginning to cut into.
AI is also expanding into supply chain visibility. By connecting data across facilities, manufacturers can track what was done to a component at a supplier site before it ever arrives at their floor, which Cutulenco called a “digital fingerprint” that travels with the part.
On Day 2, during a panel on the state of the fabrication industry featuring speakers from the steel, tooling and mechanization sectors, the conversation turned to AI as part of a broader conversation about business pressures.
Zoran Radonjic, P.Eng., of the Ontario Structural Steel Fabrication Association, drew a distinction between where AI is already delivering and where it hasn’t reached yet, separating what he called the “front room” — the shop floor and fabrication work itself — from the “backroom” of administrative work like estimating, project management and scheduling.
“I don’t see much AI being implemented in the front,” he said. “But definitely in the backroom — AI is a very, very valuable tool.”
Implementation still comes back to fundamentals
At both panels, the most consistent message was that AI projects depend on the same fundamentals as any other technology investment.
Cutulenco said companies often underestimate how much preparation is required before AI can deliver results.
Papachronis pointed to similar issues at a more technical level, noting that many manufacturers already collect data but have not validated or organized it properly.
“You have to understand… how accurate is that data,” she said, adding that companies should start with a single, well-understood use case.
Panelists also emphasized clarity of purpose. Pauly said the starting point should be defining the problem, not selecting a tool.
“When I go to visit a customer, the first question I ask them is, what are you trying to solve?” he said.
Supporting workers, not replacing them
Another theme that came up repeatedly in these discussions was how AI is being positioned inside organizations and what that means for the people who work there.
But rather than replacing workers, speakers described it as a way to support decisions, improve consistency and free up time for higher-value work.
Papachronis said companies using AI in production environments have, in some cases, increased output enough to require additional hiring.
“We’ve seen people actually increase their throughput because of the AI tools that they’ve implemented,” she said.
At the same time, speakers cautioned against relying too heavily on automated outputs.
“AI can solve problems, but cannot think on its own,” Pauly said. “Don’t take your brains out of the picture.”
In the fabrication panel, Nick Drake of Gullco International framed it more simply.
“It has to be a tool in the belt,” he said. “It’s not something that can take over and do the job for you.”
Investment decisions still taking shape
As companies move from experimentation to implementation, questions around investment and integration are becoming more prominent and the stakes are high enough that getting it wrong is costly.
Drake, whose background is in mechanized welding, said the asymmetry of a large capital commitment focuses the mind.
“You either get it right and you’re the hero, or you get it wrong,” he said.
Radonjic pointed to a specific failure mode he sees playing out as more shops begin to automate: investing in systems department by department, without accounting for how those systems need to work together.
“Those departments don’t talk to each other,” he said — a problem that compounds quickly when estimating, production and fabrication are each running on different timelines and data.
That is leading many to recommend a more incremental approach, starting with a specific use case and expanding once there is a clear return.
“Start small,” Papachronis said.
Keynote highlights pace of change
The Day 2 keynote from Amber Mac, a Canadian technology journalist and host of The AmberMac Show, addressed many of the same themes from a broader perspective, focusing on how quickly AI is evolving and what that means for organizations trying to adopt it.
“We are in the early days of the artificial intelligence revolution,” she said. “Change has never happened this fast before, and it will never, ever be this slow again.”
Her keynote landed the week after the federal government released its long-awaited AI strategy, a plan called AI for All that identifies manufacturing, healthcare and education as priority sectors. Mac acknowledged it, but said the more pressing issue for most organizations is internal: not whether to adopt AI, but how to do it in a structured way.
She outlined a three-part framework built around risks, rewards and readiness, pointing to data privacy and erosion of trust among the most immediate concerns, and noting a sharp rise in the number of organizations assigning formal responsibility for AI governance.
“Who is in charge when it comes to figuring out what these AI risks are?” she said.
On readiness, Mac was direct about where Canada stands. Only about 12 per cent of Canadian businesses are successfully adopting AI, she said, and fewer than a quarter of Canadian workers have received any AI training, well below the global average.
“Training is the biggest AI opportunity right now,” she said — a point reinforced by BDC data showing 86% satisfaction among businesses that train employees on AI, compared to 53 per cent among those that don’t.
Throughout, Mac was careful to frame AI as a tool, not a replacement — reframing the familiar “human in the loop” concept deliberately: keep humans above the loop, not just in it.
https://www.canplastics.com/canplastics/ai-drives-discussions-at-fabtech-canada/1003469588/
