Hitting the productivity bullseye: What Smart Manufacturing Week taught me about making AI work for UK manufacturers
The hall was packed—not just with the usual suspects, but with operations directors, engineering leads, and information technology (IT) decision-makers who'd clearly come with specific questions and a real appetite for answers. The conversations at our stand, in the keynote room and around the roundtable tables had a quality to them I haven't always experienced: practical, direct and genuinely curious.Here are my key takeaways.
The keynote: Where AI goes wrong—and how to get it right
I opened my session in the Manufacturing Digitalisation Summit with a provocation: The biggest barrier to artificial intelligence (AI) success in manufacturing isn't access to technology. It's knowing where to start, how to navigate a crowded landscape and how to build the organisational capability to make it stick.
We see it constantly. Manufacturers running ambitious pilots that deliver impressive demos, only to stall. Proofs of concept (POCs) that never reach production. AI tools deployed in isolation, disconnected from the workflows and operational data that actually drive decisions. The result can best be described as the productivity mirage: visible investment, invisible return.
The antidote is precision—applying the right technology to the right problem, in a structured, sequenced way. That's exactly what Infor™ Velocity Suite is built for. The methodology runs in three stages:
- Diagnose: Use process mining to surface bottlenecks, inefficiencies and high-value opportunities buried in your operational data. One electrical supply distributor we work with achieved 86% faster identification of process issues. Manufacturers using this approach are seeing up to 90% faster issue identification overall.
- Automate: Apply robotic process automation, intelligent document processing and workflow automation with precision. Not broadly. Not experimentally. But to the specific processes where the value is highest. One of our customers achieved 75% faster returns processing as a direct result.
- Optimise: Layer in machine learning, AI agents and generative AI to drive continuous improvement and competitive advantage at scale.
I walked through a detailed example: the procure-to-pay process in industrial manufacturing. It's a process most manufacturers think they understand—until you apply process intelligence to it and discover where cycle times are underperforming against industry standards, where duplicate supplier checks are creating invisible drag, and where automation can reduce process steps by 65% while freeing working capital and enabling early-payment discounts.
The point isn't just that these outcomes are achievable. It's that they're already being delivered:
☑ for an industrial conglomerate that cut auditing costs by 90%
☑ for an equipment manufacturer projecting significant time savings across procurement, supply chain, manufacturing, and sales through generative AI
Precision over ambition. Every time.
The roundtables: Candid conversations about the real state of AI adoption
I hosted four roundtable sessions on AI in manufacturing, and I'll be honest—they were the highlight of my week. The conversations were frank in a way that formal presentations rarely allow.
The picture that emerged was nuanced. Most participants were still in what I'd call the experimentation phase: running pilots, testing use cases, trying to build internal conviction. A smaller group had moved beyond that—with live deployments and measurable results. But even within that group, the journey hadn't been straightforward.
Two participants stood out for their strategic clarity, and for very different reasons.
The first was investing heavily in something unglamorous but essential: data foundations. They were building semantic layers—structured, contextual, connected data—before deploying AI at scale. It's exactly the kind of disciplined groundwork that separates manufacturers who will scale AI successfully from those who'll be back at the drawing board in 18 months.
Not exciting to present to a board, perhaps, but it is the right call—and something Infor does particularly well. Purpose-built, industry-specific applications and embedded best-practice business processes contribute to highly contextual, AI-ready data out of the box. The second had taken a genuinely process-led approach: Systematically analysing their operations to identify the specific use cases where AI would have the greatest impact on business performance—before touching a single tool. Diagnosis before deployment. It maps precisely to the Infor methodology I described in the keynote, and it was validating to see it applied independently in the real world.
The common thread across all four sessions? Manufacturers know AI matters. What they're hungry for is a trusted path from ambition to impact—not more case studies about what's theoretically possible, but a clear methodology for how to make it work in their environment, with their data, in their operational context.
That's not a gap I take lightly. It's a responsibility.
On camera and off script
I also sat down for two video interviews during the show. Both conversations focused on the same theme: What does tangible value from AI actually look like for manufacturers—and how do you get there without disrupting the operations you're trying to improve?
It's an interesting question because the answer is counterintuitive. The manufacturers who are deriving the most value from AI aren't the ones who've been boldest in their deployments. They're the ones who've been most disciplined in their targeting—who've treated AI as a precision instrument rather than a transformation programme.
Disrupt your industry, not your operations. That's not just a phrase we use at Infor—it's the principle that separates durable AI outcomes from expensive experiments.
What the atmosphere told me
There was something different about this year's show. I've been to events where the energy around AI felt performative—a lot of vendor theatre, not much operational substance. SMW 2026 wasn't that.
The question has shifted. It's no longer “Should we invest in AI?” That debate is settled. The question now—the one being asked in every serious conversation on that floor—is “How do we make it work, now, without getting burned?”
It's also a question that plays to Infor's strengths: Industry-specific solutions, embedded AI that works within your existing processes and a structured methodology—delivered through Infor Velocity Suite—that takes you from diagnosis to optimisation without the false starts.
UK manufacturing has real momentum right now. The ambition is there. The urgency is there. What's needed is precision in execution—and a partner such as Infor that understands the operational realities of your industry, not just the technology.
To explore how manufacturers are turning AI ambition into measurable outcomes, watch our on-demand webinar, AI That Knows Your Industry.