From Hype to Use Cases: What Manufacturers Need from AI

Artificial Intelligence (AI) is a hot topic in just about every corner of the world. For manufacturers, it’s gone beyond simple talking points and keynote addresses. AI has rapidly become a major focal point within strategic plans and has increasingly become a staple on the factory floor. But as manufacturers navigate this crowded and fast-moving landscape of solutions, one critical question emerges: 

What’s actually worth investing in?

That was the focus of a panel discussion at IIoT World Manufacturing Day 2025 titled “Decoding Industrial AI: What Manufacturers Need (and What They Don’t).” 

The discussion comprised leaders from Jabil (a top contract manufacturer), AWS (a premier cloud infrastructure provider), and Arch Systems (a leading provider of AI solutions for manufacturers). Together, they cut through the rhetoric and explored real industrial AI use cases that are already delivering measurable value.

“AI is both overhyped and underhyped,” said Andrew Scheuermann, CEO and co-founder of Arch Systems. “It reminds me of the dot-com boom. Everyone is slapping ‘.com’ on everything, even when it doesn’t add any value.”

He pointed to the rise of generic chatbots that don’t go beyond summarizing public content. “ChatGPT can already read your website. You haven’t really added anything by just putting a chatbot on it.”

The primary takeaway from the discussion was that simply using AI is not enough. It has to create real business value through focused AI use cases in manufacturing environments.

One of the biggest challenges to AI adoption is data integration. Research from Tech-Clarity and MESA International shows only 6% of manufacturers rate their ability to move data from collection to analysis as excellent.

Yet despite these hurdles, a growing number of companies are finding success by starting with focused, high-impact use cases in manufacturing applications. Below are several topics discussed during the panel that are already driving measurable value.

1. Predictive Maintenance and Quality

While machine learning has been around for years, it remains a powerful tool with plenty of room for broader adoption.

Scheuermann shared how supervised models trained on sensor data are being used to predict machine failures before they happen. Similar approaches are helping manufacturers identify quality issues earlier in the process, reducing scrap and downtime.

This is one of the most widely adopted AI use cases in manufacturing today. 

2. AI-Powered Root Cause Analysis

Generative AI is especially well-suited for root cause analysis. It can quickly interpret a wide range of unstructured data, including dashboards, logs, and PDFs.

“Today, when there’s a downtime or quality issue, people run around reading reports and checking screens,” Scheuermann said. “Now AI can do it in seconds. It can find the needle in the haystack.”

This is among the most high-impact AI use cases in manufacturing, especially in complex operations with recurring quality challenges.

3. Internal Knowledge Assistants

Doug Bellin, Global Smart Factory Lead at AWS, encouraged manufacturers to start with use cases that improve access to internal knowledge.

“Think about all the manuals, SOPs, and troubleshooting guides locked in SharePoint or a file cabinet,” he said. “With generative AI, you can make that data searchable in natural language. That saves time and reduces mistakes.”

These tools can act as AI copilots for technicians, offering real-time answers drawn from a company’s own documentation which is an emerging AI use case in manufacturing knowledge transfer.

4. Guided Actions and Decision Automation

The next level is not just understanding the problem but guiding or automating the next step.

Bellin gave the example of a machine error code. Instead of forcing a technician to look up a solution, the system could immediately suggest a repair method or share a video tutorial based on past fixes.

Scheuermann referred to this capability as “agentic AI,” where the system can reason through uncertainty and support or even initiate human-like actions.

These are fast-growing AI use cases in manufacturing automation.

5. Augmented Shift Reports

Another practical use case is automating daily production reports.

“Some companies spend hours building shift reports,” said Bellin. “AI can pull that data instantly and generate a complete summary of OEE, inventory, and bottlenecks. It can even suggest what to do next to stay on plan.”

This not only saves time but also improves consistency and transparency across teams, making it one of the most impactful AI use cases in manufacturing operations.

May Yap, SVP and CIO at Jabil, emphasized the importance of aligning people, processes, and systems from the very beginning.

“It doesn’t matter whether the initiative is led by IT or OT,” she said. “What matters is having a champion who brings everyone together, including business units.”

She also stressed that every AI project at Jabil is tied to a clear business case. That includes measurable KPIs such as first-pass yield and downtime reduction.

“Start with the question, what problem are we solving? Then ask, how does this improve performance, and where will we see ROI?”

The panel was optimistic about the accessibility of these technologies, even for small and mid-sized manufacturers.

“You don’t need a massive data lake or your own large language model,” said Scheuermann. “In some cases, you don’t even need to connect directly to a machine. AI can read the screen like a human and act on what it sees.”

Bellin added, “Start with one use case. One machine. One dataset. You’ll learn quickly and see results faster than you might expect.”

For manufacturers, the takeaway is clear. Focus on solving real problems, not following trends. The companies seeing the greatest success with AI are those aligning their efforts with business goals, not just experimenting with flashy tools.

Whether it’s automating root cause analysis, enhancing frontline support, or eliminating routine data entry, industrial AI use cases are already producing real impact.

The difference is knowing where to look and having the right strategy to get started.