March 31, 2026
Where Do We Start with AI? Start with the Business Problem—Not the Technology
Why most AI projects fail before they begin and how manufacturers can get it right
Manufacturers are feeling the pressure to act on AI. The conversation is everywhere—from the shop floor to the leadership team—and many manufacturers are asking, “Where do we start?”. According to Sanjay Mohan, Executive Director of AI Strategy with the MKE Tech Hub Coalition’s Synapse Initiative, the biggest risk isn’t waiting too long—it’s taking the wrong first step.
“Fewer than 1% of AI projects reach production with measurable return on investment,” Sanjay says. “That’s not because AI doesn’t work. It’s because most organizations start in the wrong place.” And when that first step goes wrong, the impact can last much longer than the pilot itself.
A Bad First Experience Can Shut the Door
When AI efforts don’t deliver, the consequences go beyond wasted time or budget. Sanjay sees this play out in a familiar way. A team tests a generative AI tool on something it wasn’t designed for—like production estimates or process analysis. The results are inconsistent. Confidence drops quickly. “They try it once, it doesn’t work the way they expected, and the conclusion is: ‘AI doesn’t work,’” he explains.
But the issue isn’t the technology—it’s how it was used. “Generative AI is probabilistic by design,” Sanjay says. “It’s great for drafting content. It’s not built for process control.” Still, the takeaway across the organization is the same: AI didn’t deliver. Teams become skeptical. Leadership becomes hesitant. Momentum stalls. “One bad first experience can poison the well,” Sanjay says. “Not just for AI—but for technology investment broadly.”
How Good Intentions Lead to the Wrong First Step
Most companies don’t get here by accident. In fact, the path often starts with good intentions. An employee experiments with a tool. They see potential. They bring it forward. Leadership wants to support innovation and approves a pilot. “There’s real value in that bottom-up enthusiasm,” Sanjay says. “It creates engagement you can’t get from a top-down mandate.” The problem is that the enthusiasm is based on a limited view of what AI can do.
Most teams are only exposed to generative AI tools. They’re not seeing the broader landscape—machine learning, computer vision, predictive analytics, process optimization. So the ideas that surface reflect what people know. “The pilot gets approved,” Sanjay explains. “But it’s not connected to the company’s real challenges. So it doesn’t move the needle.” The result is frustration—and hesitation to try again. “That’s not a technology failure,” he says. “It’s a strategy failure.”
The Right First Step: Start with the Problem
Strong AI strategies don’t begin with tools. They begin with the business. “The first step isn’t choosing a tool,” Sanjay says. “It’s asking: what are our biggest challenges—and what’s causing them?”
That means stepping back and looking at operations with fresh eyes:
- Where are we losing time?
- Where does quality break down?
- Where are we leaving money on the table?
- What’s keeping leaders stuck in firefighting?
From there, the focus shifts to root causes. A slow quoting process might look like a technology issue. But the real problem could be disconnected systems, inconsistent practices, or knowledge that only one person holds. “Understanding the root cause changes everything,” Sanjay explains. “It determines whether AI is even the right solution.” Sometimes it is. Sometimes it isn’t. The key is making that decision from a position of clarity—not urgency.
Clarity Turns Fear into Confidence
Right now, many manufacturers are navigating uncertainty around AI. They don’t want to fall behind. They don’t want to make the wrong investment. And at the same time, they’re facing workforce challenges—experienced employees retiring and taking critical knowledge with them. “That fear is real,” Sanjay says. “But it’s not a strategy.”
What changes the conversation is structure. “When manufacturers step back, assess their business, and prioritize the right problems, the fear starts to go away,” he explains. “Not because they’ve implemented anything yet—but because they finally have a map.” That clarity gives teams direction—and confidence. “They realize they’re not behind,” Sanjay says. “They’re getting ahead.”
Bottom Line: Start with a Plan, Not a Pilot
For manufacturers exploring AI, the takeaway is straightforward. Don’t start with the tool. Start with the problem. Understand what’s holding your business back. Identify the root causes. Then evaluate where AI fits—and where it doesn’t. Because the goal isn’t to try AI. It’s to solve real problems and improve performance. “The best first step isn’t the fastest one,” Sanjay says. “It’s the right one.”