Agents are taking over work that slows operations. Companies that harness them effectively will pull ahead.
A recent Verdantix model predicts the industrial AI analytics software market will grow from $3.2 billion in 2025 to almost $9.3 billion by 2031. This rapid growth, about 20% per year, reflects a real-world need. Traditional software has hit a ceiling: it helps us spot problems, but not solve them. Now, we’re seeing tools that don’t just highlight issues; they can take action, too.
From Information to Action: Bridging the “Agency Gap”
The biggest friction point in modern operations is what I call the Agency Gap, the distance between receiving a piece of information and completing the task.
Most industrial software today is “Knowledge Rich” but “Action Poor.” Dashboards tell you what happened, and Machine Learning gives you predictions. Both still require a human to “bridge the gap” between insights and action. It becomes an inefficient, multi-step process to identify the root cause, create a work order, check inventory, and email a vendor to check the order status.
An Industrial AI workforce takes action on what it learns as an agentic system. It interprets machine signals, automates decision-making, and seamlessly moves work between systems. As software evolves from reporting and recommending to enabling action, agent-led operations free teams to focus on solving real problems instead of routine tasks.
The Persona-First Framework
To build a valuable agent, you must begin with the Persona. By focusing on the persona’s “Jobs to be Done” (JTBD) and the specific roadblocks of their team, you ensure the technology meets business needs rather than forcing teams to adapt to the tech.
Before you write a single line of code, map the human experience through five lenses:
- Start with the persona: A maintenance planner, plant manager, or logistics lead, each sees different challenges and opportunities, and your approach should reflect that.
- Map workflows and Jobs to Be Done: Understand their daily work, key decisions, and gaps.
- Audit the Tools: Which siloed data sources (CMMS, Historian, MES, ERP) do they touch?
- Identify “Swivel Chair” Operations: The more systems a person must jump between to complete one task, the more ‘giddy’ the person feels and the more powerful an agent becomes.
- Quantify Productivity: If the agent handles the “swivel chair” tasks, how much capacity is returned to the human?
Persona-first agents unlock time and capacity, giving teams a data-driven agenda at the start of each shift focused on the highest-value tasks for the business. The time freed for human experts becomes strategic fuel. They can tackle complex problems, mentor colleagues, and push continuous improvement forward. Less time spent on routine coordination means more focus on innovation, safety, and initiatives that actually move the organization.
The “Swivel Chair” in Action
Map the moments where work slows down or waits on human coordination. Agents thrive in those gaps, turning friction into speed and impact.
| Persona | The Workflow | The “Swivel Chair” Operation |
| Maintenance Planner | Parts ordering & scheduling | Jumping between Predictive Dashboards to identify an alert, checking the CMMS for technician availability and asset history, and then verifying part lead times in the ERP. |
| Plant Manager | Performance & safety reporting | Aggregating data from Historian dashboards and MES production logs into Excel to reconcile downtime events for the morning production meeting. |
| Logistics Lead | Shipment tracking | Toggling between Carrier Portals for real-time status, the WMS/ERP for order details, and Teams/Email to coordinate with production about delayed raw materials. |
| Supply & Demand Planner | S&OP (Sales & Operations Planning) Alignment | Manually reconciling CRM (Salesforce) demand spikes with ERP safety stock levels and real-time MES production logs to confirm if the floor can fulfill top-line commitments. |
Teams across industries often spend more time connecting disconnected data than using the insights it offers. Agents bridge these gaps, eliminating manual “swivel” work and adding hours of productivity that create value for workers and their companies.
Example: AI Agents in Industrial Maintenance
Consider a technician who receives a vibration alert on a critical pump.
Today: Technicians jump from one system to the next, pulling logs, checking parts, building work orders, and chasing approvals. Every handoff slows the process and creates room for mistakes.
The Agent-Led Future: The agent catches the alert, verifies parts, and queues the work order. The technician sees one simple prompt: “Pump 4 bearing needs replacement. Parts are staged. Approve work order?” Tasks that used to waste time now happen in seconds, letting the team focus on the problems that truly require their expertise.
Why Industrial is the Ultimate Use Case
While agentic AI is taking off in the front office, it is revolutionary for the factory floor because of four structural advantages unique to industrial environments:
- The Expertise Multiplier: Experienced operators have a “feel” for the plant.They recognize a machine’s “hum” before it fails. Agents codify that intuition. Trained on the patterns veteran technicians know instinctively, they automatically trigger the right maintenance flow and the right time, scaling irreplaceable human expertise across every shift at every site.
- The Integration Opportunity: Industrial data is notoriously siloed. But that complexity favors organizations already invested in the space. Agents act as the “intelligent glue” that turns fragmentation from a liability into a competitive edge.
- The Domain Context Edge: Unlike generic LLMs, industrial agents understand the physics of a production line and the nuances of a Programmable Logic Controller (PLC). This is context that can’t be improvised and takes years to develop.
- The Data Compounding Advantage: Agents are only as smart as the data they learn from. Organizations with years of proprietary operational data, such as equipment history, failure patterns, and maintenance outcomes, have a structural head start that only grows stronger over time.
The next chapter for Industrial AI lives on the factory floor, where the data is the deepest, the stakes are highest, and the gap between knowing and doing hold the most value yet to be unlocked.
The New Math: Unlocking $10B+ in Economic Productivity
When I talk about a $10 billion impact, I usually get one of two reactions: a blank stare or a roll of the eyes. It sounds like one of those massive, theoretical numbers marketing teams use to fill space.
It’s actually a very conservative look at the technical debt we’re forcing our people to pay every single day.
Look at the specialists running operations from the senior engineers and operators to maintenance leads who actually keep the lights on. There are about a million of these high level members in the US industrial workforce. On average, they’re spending at least 30% of their day on what I call “digital chores”, manually syncing data between systems, chasing down status updates, and fighting with clunky software interfaces.
That’s where the math gets interesting.
If we use agents to automate those tasks and give that 30% of their time back, we aren’t just making them “more efficient.” That’s economic value and productivity added.
Even if you’re skeptical and think we’ll only capture half of that potential, you’re still looking at $10 billion in economic productivity every year.
This isn’t about cutting heads or replacing humans. It’s about a “Capacity Dividend”. It’s about finally letting your most expensive, most talented people stop acting like data entry clerks and start doing the high-value work you actually hired them for. That $10 billion is just the floor. When those experts are finally free to focus on things like yield and uptime, the real world ripple effect on the bottom line is many times larger.
The Capacity Dividend: Augmentation, Not Replacement
A common concern is that “Agents” means “fewer people.” We must address this misunderstanding head-on: The goal of these agents is not to replace people, but to promote them.
Manufacturing is currently facing a massive talent shortage. We don’t have enough people to do the work that is already required. By bringing a 30% increase in productivity, we are creating a Capacity Dividend.
This dividend allows your team to finally tackle the “Backlog of Excellence”, the strategic, high-value projects they are currently too busy to touch. We are moving humans from “Data Clerk” to “Decision Maker.”
Building the Trust Bridge: Human-in-the-Loop
Moving from insight to action requires trust. In heavy industry, we cannot simply “turn on” autonomy and walk away. The first phase of the Agent-Led Revolution is built on Human-in-the-Loop workflows.
An agent identifies a bearing failure, verifies parts availability in the ERP, and drafts the work order, but a human provides the final “okay” before the system executes. This keeps the accountability with the experts while removing the administrative friction that slows them down.
What’s Next: Closing the Gap Between Knowing and Doing
The “swivel chair” is slowing your operations down. For years, the sheer volume of data and fragmented systems made managers feel “giddy” with overwhelm. But by closing the Agency Gap, we turn that around.Take a look at your own day: What is the most frustrating “swivel chair” operation you perform? That is exactly where your first agent lives. It’s time to stop just “knowing” and start “doing.” That is when things truly get “giddy” in the best way possible.
Key Takeaways for Executives:
- Focus on the Gap: Solve for the distance between information and action.
- Start with the Human: Map the “swivel chair” tasks of your most burdened personas.
- Measure Outcomes: Move the conversation from software costs to realized business value.
FAQs
How does agentic AI change operational decision-making?
Agentic AI moves beyond alerts and insights, enabling systems to act on data and coordinate across teams and machines. This shifts how decisions are made and where human attention is focused.
Why should executives care about “swivel-chair” tasks?
These repetitive cross-system tasks reduce productivity and obscure strategic priorities. Addressing them highlights where automation can deliver immediate operational value.
Where should we start when implementing agentic AI?
Focus on the workflows that carry the highest operational friction or risk. Starting small in high-impact areas builds trust and demonstrates measurable results quickly.
How do we measure the value of agentic AI adoption?
Look at outcomes such as reduced downtime, faster response to anomalies, improved throughput, and the time freed for teams to focus on higher-value initiatives.
What are the risks of delaying adoption?
Organizations that resist agentic AI may miss opportunities to optimize operations, reduce costs, and maintain competitive advantage as peers leverage autonomous decision-making.
