How Manufacturers Are Unlocking Operational ROI Through Agentic Workflows

Move from fragmented IT/OT data to hard-dollar P&L impact

It’s 2 a.m. on a Tuesday. Deep inside a Midwest manufacturing plant, a critical stamping press throws a fault code.

The maintenance team scrambles. But the data that could have predicted this exact failure three days ago is trapped in a legacy system that hasn’t been updated since 2014 and doesn't talk to your ERP or scheduling engine. By the time leadership is briefed at morning standups, a full shift of production is lost, a major customer’s order is at risk and supply chain teams are frantically dialing backup suppliers off an outdated spreadsheet.

If this sounds painfully familiar, you’re far from alone.

While the tech world obsessively debates how generative AI can draft marketing copy or summarize email threads, industrial leaders are fighting a vastly different battle. For manufacturing executives, artificial intelligence isn't about soft-dollar administrative savings — it's about operational survival, P&L returns and physical safety.

According to findings from the Global AI & Data Trends Survey (a study of 2,050 global enterprise leaders, including nearly 300 manufacturing decision-makers), the manufacturing sector is navigating a unique AI paradox.

Here’s a preview of some of the impactful findings the data revealed, including where manufacturing really stands with AI, the hidden roadblocks holding back autonomy and why the industrial sector might actually be winning the real AI race.

The deployment paradox: selective, pragmatic and leading the pack

There’s a persistent myth that manufacturing trails other industries in digital maturity. On paper, that claim seems plausible: 41% of manufacturing firms report being in the “initial use cases” phase, slightly lower overall maturity than financial services or media.

But if you dig deeper into where AI is actually deployed, a significantly different narrative emerges.

• The operational gap: 52% of manufacturing organizations have live AI running in core supply chain and operations, compared to just 36% across all other industries.

• Supporting critical business needs: Manufacturers aren’t interested in generic AI assistants or merely eliminating rote tasks. They are prioritizing operational efficiency (57%), product innovation (48%) and operational R&D acceleration (59%).

While other sectors chase broad adoption metrics by deploying tools that yield incremental savings, manufacturers are playing a higher-stakes, higher-reward game. They are embedding intelligence directly into their core operational engines to slash downtime, prevent stockouts and squeeze maximum margin out of every asset on the plant floor.

The great workforce reshuffling: upskilling over automation

The conversation around AI and jobs is often reduced to simple alarmism: Automation is coming for your workforce. But the data shows manufacturing is undergoing a structural transformation, not merely a reduction in force.

Manufacturing is experiencing the highest simultaneous rates of job churn of any sector:

• 31% of manufacturers report jobs lost within operations due to AI.

• 32% report jobs gained on those same teams.

• Net-positive reality: Across all enterprise organizations surveyed, 77% have experienced active job creation due to AI implementation.

While we know these shifts are happening, we also know where they’re happening. The losses are overwhelmingly concentrated at the entry level (62% of job losses), where routine, manual data-handling tasks are being automated. In comparison, veteran domain experts are more secure; job losses among senior individual contributors were significantly lower in manufacturing than in other industries (33% vs. 41% overall).

The strategic framing for this change is clear: Your best operators aren't being replaced — they're being promoted to supervise the AI systems that automate what they used to do manually. Legacy roles centered around manual data entry are evolving into high-value positions like supply chain prompt engineers, digital twin analysts and agent operations (AgentOps) engineers. By leveraging domain expertise to govern autonomous systems rather than execute manual tasks, these reskilled professionals drive faster decision-making, prevent costly downtime and protect margins.

The agentic frontier: cautious exploration, massive returns

The next frontier of industrial technology is agentic AI: systems that don’t just analyze data or suggest a path, but autonomously take execution steps. In manufacturing, that might look like agents that dynamically reroute supply chains, auto-trigger calibration interventions before a press fails or negotiate spot-buy orders.

Here, manufacturers display a fascinating dual mindset:

• The caution: Compared to the other industries in the survey, manufacturing leaders are the most cautious about agentic AI, with 25% describing their stance as “cautious exploration.”

• The ROI: Despite that caution, 22% believe manufacturing will derive the largest economic return from agentic AI of any industry — more than double the cross-industry average of 10%.

• The velocity: 51% expect to have active AI agents in production within the next 12 months.

Despite the fact that manufacturers fully understand the potential benefits they stand to gain from embracing agentic AI, they have very good reason for being cautious about jumping in with both feet. In this industry, the cost of an AI error or hallucination isn't a slightly annoyed chatbot user — it could be a multimillion-dollar machine failure, a ruined batch or a severe safety incident. Manufacturing leaders aren't skeptics, but rather pragmatists who deeply understand the stakes.

The hidden roadblock: trapped, fragmented data

An abundance of caution isn’t the only thing keeping manufacturers from fully scaling AI. But it’s not the complexity of the AI models that’s posing a problem. It’s the data architecture.

• 43% of manufacturers identify fragmented data architecture as their single largest hurdle to scaling AI.

• 67% of manufacturing data still lives in disconnected databases, spreadsheets and CSVs, 11 percentage points higher than the global average.

• 54% of manufacturers rely heavily on unstructured visual data (like IoT inspection camera feeds) that legacy systems simply cannot reconcile with ERP or OT databases.

Your AI is only as good as your data. If an AI agent can’t quickly connect a raw material shipping delay in Asia to a sudden temperature spike on a machine in Ohio, it can’t act to help save your production shift. Unifying IT, OT and IoT telemetry into a single, governed data layer isn't just a nice-to-have IT project; it’s the critical prerequisite for achieving industrial autonomy.

Getting from pilot purgatory to autonomous operations

Navigating this complex terrain requires more than just implementing new software; it also requires a structured, multi-stage roadmap that connects the realities of day-to-day operations directly to your business P&L.

Whether you need to bridge the gap between legacy systems and modern data infrastructure, establish strict governance that accelerates innovation or pitch C-suite leadership on funding your next AI initiative, you need a playbook grounded in real-world industrial execution.

In our ebook “Navigating the Age of Agentic AI: A Manufacturing Leader’s Playbook,” you’ll discover:

• Real-world use cases: How industry leaders successfully scaled AI while avoiding pilot purgatory.

• The four-stage operational roadmap: A step-by-step guide to moving from fragmented data modernization to full multi-agent orchestration.

• The C-suite framework: Exactly how to pitch AI investments to CFOs by focusing on the issues that matter to them, such as hard P&L metrics like downtime reduction and recovered throughput, rather than tech jargon.

Download the full ebook to unlock the stats, insights and strategies you need to lead your organization into the era of agentic manufacturing.