Let me say something that may raise a few eyebrows with all the noise we are hearing about AI these days: people don’t buy “AI” or “agents”. They buy fixes to process problems.

That line has become the shorthand for everything my team believes about how manufacturers should approach AI, and I want to use this post to unpack it properly, not as a recap of the keynote, but as an actual playbook. In Chicago at our Champions of Manufacturing event we announced our vision for the future: Manufacturing Intelligence. 

Manufacturing Intelligence is QAD | Redzone’s vision for bringing manufacturing-specific intelligence across the systems, workflows and decisions that run a manufacturing business, from the shop floor to the top floor. Grounded in more than 50 years of manufacturing expertise, it helps manufacturers move from systems that simply record what happened to systems that can increasingly see what is happening, understand why, decide what to do, act and prove the outcome. It brings together AI, manufacturing data, context and know-how across the broader manufacturing environment, including systems beyond QAD | Redzone, to help people make better decisions, act faster and continuously improve business performance.

Start with the Business Problem, Not the Technology

Every AI conversation I have with a manufacturer goes sideways the moment it starts with the technology. “We want to do something with agents.” “Our board asked what our AI strategy is.” “A vendor showed us a demo and we don’t want to fall behind.”

None of that tells you where value actually lives in your business.

The manufacturers who get real traction start somewhere much more boring, and much more useful: a process that everyone already knows is broken. Not broken in the abstract, but broken in the specific, measurable way that shows up every week as a delay, a rework, a missed discount, a frustrated team. Procure-to-pay that takes too long. A quote-to-cash cycle full of manual handoffs. A sourcing process where nobody can say with confidence where time and money are actually going.

The entry point is “what is this process costing us today.”

Find the Workflow Friction

Once you’ve named the process, the next step is to actually go look at it. Not the org chart version of the process, the real one, with the people who run it every day.

We do this with customers through what we call a process re-engineering workshop, and it’s a lot less exotic than it sounds. It’s a few hours in a room with the people who own the process, mapping the current state end to end: where the handoffs happen, where approvals stall, where the same information gets re-entered three times, where good people are spending their day on work that a system should be doing for them.

The point isn’t to produce a nicer flowchart. It’s to find where time is leaking out of the process and why.

We ran this workshop with ITW on their procure-to-pay process. Their team walked in knowing something wasn’t working efficiently, which is usually true everywhere I go. What they didn’t have was a precise picture of where the friction actually lived or what it was costing them. That’s what the workshop produced.

Quantify the Value

This is the step most companies skip, and it’s the one that actually gets a project funded.

It’s not enough to know a process is slow. You have to be able to say what slow is costing you in dollars, in working capital, in missed opportunity, and what a redesigned, AI-enabled version of that process would be worth instead. Without that number, “we should look into AI” stays a conversation. With it, it becomes a business case.

Working through ITW’s procure-to-pay process this way surfaced roughly $1.2 million in annual savings and $550,000 in working-capital release, built from their own process data, with their own team in the room, not a projection from a slide. That’s the difference between an AI pitch and a business case a CFO will actually fund.

This is also where I’d push back on how most of the industry frames “getting started with AI.” A lot of what’s being sold right now assumes you need a six-month discovery engagement, a large consulting exercise, or a big technology commitment before you find out whether the value is even there. AI can now help capture the process knowledge, map the workflow, expose where time is leaking, and quantify the opportunity in a fraction of that time. Instead of committing to a transformation program on faith, you need to be looking at a quantified case before you decide anything.

Prove It

Once you have the number, the workshop leaves you with three concrete things, not just a recommendation:

• A redesigned process flow: what procure-to-pay, quote-to-cash, or source-to-contract looks like once an agent is doing the heavy lifting at each bottleneck, not bolted onto the old way of working
• A quantified business case: the ROI case for making the change, built from your own numbers
• The actual configuration for the agent that runs the new process

That last point matters more than it might seem. It’s a working starting point your team can act on immediately, internally, with your own leadership, using your own language and your own numbers, not ours.

That’s exactly what ITW’s procurement team did next: they took what came out of the workshop back to their own management and made the case themselves. The people who run the process believed in it enough to bring it to their own leadership team, which is a better proof point than anything we could say about the methodology ourselves.

Expand from There

This isn’t a strategy for one process, and it’s not a one-time event.

You don’t need an enterprise-wide AI roadmap to start. You need one meaningful workflow, a quantified result, and the confidence that comes from having proven it once. Procurement this quarter doesn’t obligate you to sourcing next quarter, but once your team has seen a process get faster, cheaper, and less frustrating to run, they usually want to know what’s next on the list.

This approach also isn’t limited to any one part of your business or any one system. Whether you’re modernizing your ERP backbone, transforming frontline execution, or tackling a workflow in procurement, sourcing, or global trade, the sequence is the same: find the friction, quantify it, prove the fix, expand. There’s no required order and no requirement to rip out what you already have to start.

Why This Has to Be Built for Manufacturing

This approach only works if the intelligence underneath it actually understands manufacturing.

Procure-to-pay, quote-to-cash, source-to-contract. These aren’t generic business processes with manufacturing labels stuck on them. They carry the specific logic, exceptions, and constraints of how manufacturers actually run their business, connected to the ERP as both the record of truth and the system where the work actually happens. That’s why ChampionAI is grounded in manufacturing expertise and connected to real manufacturing workflows, rather than built as a general-purpose layer applied to manufacturing after the fact. It’s also why governance, auditability, and human oversight aren’t an afterthought here. The goal was never autonomous technology for its own sake. It’s putting intelligence to work inside the processes where your business runs, with the people who run it still firmly in control.

The Takeaway

You don’t need a transformation program to get started with AI. You need one process that’s costing you real money today, a clear-eyed look at where the friction actually lives, a number you can defend in front of your CFO, and the confidence that comes from proving it once.

Have a process like ITW’s? Find out what it’s costing you. Learn more about ChampionAI and get in touch with our team.