Centric Manufacturing Expert Explains How to Pick an AI Pilot That Can Prove Its Value

Rob Williams, Centric Consulting’s supply chain lead, authored an article in Crain’s Cleveland Business on how manufacturers can choose AI pilots that prove their value on the production floor.

In the piece, “Your next AI pilot doesn’t have to solve your biggest problem,” Williams advises manufacturers to choose an AI use case tied to a production problem that leadership already tracks.

A pilot aimed at a metric the operation doesn’t prioritize makes success hard to prove. In a 2026 RSM survey, 30 percent of middle-market manufacturers whose AI pilots saw moderate or limited success named unclear ROI as a top reason. Tying a pilot to a KPI the business already measures gives teams a starting point for measuring the return, he said.

Data readiness is a second hurdle. In a Manufacturing Leadership Council survey, nearly half of manufacturers rated their operational data as having low AI readiness. On the production floor, some companies must first build a data foundation.

“If the data you need isn’t being captured, sometimes the honest starting point is a clipboard and a spreadsheet, tracking operations hour by hour until there is something to build on,” he said.

Williams drew on his own years as a production supervisor. After 12-hour night shifts, he spent another 30 minutes documenting production, scrap, and machine issues, plus notes for the incoming shift.

“Today, that’s the kind of job I would consider for an AI pilot,” he said.

Once supervisors capture that information consistently, Williams continued, AI can write the first draft of the report to give them that time back.

Connecting systems, building the data model, and training an AI agent typically takes about 12 weeks, which is why he recommends keeping the scope manageable.

“The first AI-enabled win doesn’t have to be the biggest one; it just has to give you a reason to build the next,” he said.

Read the full article in Crain’s Cleveland Business.