Pieter Van Schalkwyk
CEO at XMPRO
Will your organization lead the AI transformation or struggle to catch up later?
That’s the question facing industrial operations today. McKinsey’s recent “The agentic organization: Contours of the next paradigm for the AI era” report describes “the agentic organization” as the largest shift since the industrial revolution. This isn’t hype. We’re watching a genuine phase transition in how humans and machines coordinate to create value.
What makes this relevant for industrial operations? XMPro MAGS with APEX already demonstrates this transformation today. This three-part series provides the concrete map for what McKinsey describes in strategic terms.
The Core Insight: Execution vs. Orchestration
Understanding the agentic transition requires grasping one fundamental distinction. In traditional operations, humans execute work and coordinate with each other. In agentic operations, AI agents execute work while humans orchestrate agent networks. This sounds simple, but the implications cascade through every aspect of organizational structure.
Consider reliability management in a refinery. The traditional model assigns reliability engineers to monitor equipment health. Maintenance planners schedule interventions. Operators adjust production around maintenance windows. Supervisors coordinate across these functions. Each role involves substantial execution work: analyzing vibration data, creating work orders, adjusting process parameters, and conducting meetings. Coordination overhead often exceeds actual value creation.
The Agentic Model Inverts This Structure
Specialized agents continuously monitor equipment streams. They detect anomalies by pattern matching against historical failures. They cross-reference with maintenance records to identify intervention timing. They coordinate with production scheduling agents to optimize downtime windows. They generate work orders with specific parts and procedures. They execute routine responses autonomously.
The human reliability team (now perhaps two to three people instead of a dozen) oversees this agent network. They handle exceptions that fall outside agent capabilities. They make strategic trade-offs when objectives conflict. They continuously improve agent behavior based on operational learning.
This isn’t automation in the traditional sense. Automation replaces human execution with programmed logic. Agentic systems replace human execution with adaptive intelligence that learns, reasons, and coordinates. Automation breaks when conditions change. Agentic systems adapt to novel situations within learned boundaries.
Business Model Transformation: Three Critical Shifts
McKinsey identifies three business model shifts that define agentic advantage:
- AI-first workflows drive marginal costs toward compute costs rather than coordination costs
- AI-native channels enable real-time personalization, impossible with human-mediated interactions
- Proprietary data becomes the decisive competitive differentiator
XMPro MAGS demonstrates all three principles in industrial contexts.
Shift 1: Solving the Skills Crisis While Transforming Economics
The transformation addresses a critical challenge facing industrial operations: you cannot find the skilled people you need. Experienced operators are retiring faster than you can train replacements. Process engineers with deep domain knowledge are scarce. The talent pipeline cannot keep pace with operational demands.
Consider the reality of continuous operations. Traditional models require three to four operators per shift, plus engineers on call. That’s 12 to 16 skilled people just to maintain 24/7 coverage for one unit. Multiply this across multiple units and you face an impossible staffing challenge. Where do you find these people? How do you train them quickly enough? What happens when your most experienced operators retire?
With MAGS, specialized agent teams conduct continuous assessment cycles at the same rate, or better, than humans do. They respond to deviations within minutes. They escalate to human operators only when conditions exceed defined complexity thresholds. The facility operates autonomously for days or weeks between human interventions.
This isn’t about eliminating jobs. This is about operating sustainably when the skilled workforce you need doesn’t exist. Your existing skilled operators become supervisors overseeing multiple units instead of monitoring single control rooms. You’re multiplying the impact of scarce expertise rather than trying to replace it.
Here’s where the economics become transformative. Traditional operations have marginal costs tied to human coordination and labor. Adding another unit means adding another full shift rotation. AI-first workflows fundamentally change this equation. The marginal cost of monitoring additional units approaches the cost of compute rather than the cost of hiring and coordinating more people. One supervisor with agent support can oversee what previously required multiple shift teams. This solves both the availability problem (you don’t have the people) and the economics problem (you couldn’t afford them if you did).
Shift 2: Real-Time Capabilities
Real-time capabilities emerge from agent-to-agent coordination at machine speed. Traditional operations involve handoffs. A process upset triggers an alarm. An operator assesses the situation. They contact the supervisor or engineer for guidance. They implement a response. They document the action. This sequence requires fifteen to thirty minutes minimum.
Agent teams eliminate handoffs. Assessment, coordination, and response happen simultaneously through parallel agent processing. This isn’t faster humans. This is a qualitatively different operational mode where sequential handoffs become obsolete.
Shift 3: Proprietary Data Advantage
The proprietary data advantage builds over time as you accumulate operational intelligence. XMPro DataStreams connects to over 150 industrial systems. These include SCADA, DCS, historians, maintenance management, and quality systems. We continuously capture operational context. Agents “remember” context better than humans do.
The agents learn from this context. Which operator responses proved effective for specific upsets? What maintenance interventions prevent particular failure modes? How do process interactions create cascade effects under unusual conditions? This knowledge compounds with operational experience. It creates a competitive moat that others cannot easily replicate. Done right, it is unique, scalable, and difficult for your competitors to emulate.
Operating Model: From Hierarchies to Networks
McKinsey describes agentic organizations as flat networks of outcome-focused teams. This contrasts sharply with functional hierarchies or even cross-functional digital teams. The structural change is more radical than it first appears.
Why Traditional Structures Exist
Functional hierarchies organize around specialized expertise. Process engineers report to the chief engineer. Maintenance technicians report to the maintenance manager. Operators report to the operations supervisor. This structure made sense when coordination happened through formal channels (meetings, work orders, approval chains). It worked when execution required specialized human skills.
Digital cross-functional teams improved on this by embedding diverse expertise within teams. But they still assumed human execution and limited team size (the “two-pizza team” constraint).
Agentic Teams Transcend Both Models
Agentic teams organize around outcomes rather than functions or products. Agent execution eliminates team size constraints. A small group of two to five humans with combined expertise can supervise an agent network. That network might include 50 to 100 specialized agents delivering end-to-end outcomes.
The human team provides strategic direction. They handle complex exceptions. They continuously refine agent behavior. The agent network executes everything else.
How XMPro MAGS Implements This
XMPro MAGS implements this through specialized agent teams combining technical analysis, coordination, and regulatory oversight. A reliability agent team might include the following:
- Equipment monitoring agents that process sensor streams
- Diagnostic agents that identify failure modes
- Coordination agents that optimize maintenance timing against production schedules
- Compliance agents that ensure regulatory requirements
- Supervisor agents that provide continuous oversight of the entire network
The human reliability team doesn’t execute any of these functions directly. Instead, they define the objective functions that guide agent behavior. They maximize equipment availability while minimizing maintenance cost and ensuring safety compliance. They monitor performance against these objectives. They adjust agent parameters when performance drifts. They handle situations that require human judgment.
Rethinking Organizational Boundaries
This structure requires rethinking organizational boundaries. Traditional organizations draw clear lines. This team owns production. That team owns maintenance. Another team owns quality.
Agentic teams organize around outcomes that span these traditional boundaries. The reliability team owns asset performance. This requires influencing production schedules, maintenance execution, and operating procedures simultaneously. The agents coordinate across these domains automatically. They eliminate the coordination overhead that plagues traditional matrix structures.
This does require a rethink of your business processes. We are no longer trying to digitize the processes that originated when we still moved paper around in brown envelopes with small holes.
Human in the loop processes 🙂
We can now create dynamic processes that continuously optimize towards real-time monitored objective functions.
What This Means for Your Organization
McKinsey notes that 89 percent of organizations still operate with industrial-era hierarchies. Nine percent have adopted digital-era agile models. Only one percent operate as decentralized networks (the organizational structure most compatible with agentic operations).
This distribution suggests enormous headroom for competitive advantage through earlier adoption. It also reveals that the transition is genuinely difficult. Most organizations remain in older structures not because they lack awareness. They remain there because the transformation challenges are substantial.
The question facing industrial leaders isn’t whether agentic operations will arrive. The choice is whether they will lead the transformation or struggle to catch up after competitive disadvantage becomes undeniable.
Part 2
In the next article, we’ll explore the critical governance and workforce transformations required for agentic operations.s. You’ll learn:
- How to implement real-time governance that operates at machine speed
- The three emerging workforce roles (M-shaped supervisors, T-shaped experts, AI-augmented frontline workers)
- Practical strategies for workforce transition and capability development
The industrial future is agentic. The only question is whether your organization will help create that future or be disrupted by it.
Pieter van Schalkwyk is the CEO of XMPro, specializing in industrial AI agent orchestration and governance. Drawing on 30+ years of experience in industrial automation, he helps organizations implement practical AI solutions that deliver measurable business outcomes while ensuring responsible AI deployment at scale.
About XMPro: We help industrial companies automate complex operational decisions. Our cognitive agents learn from your experts and keep improving, ensuring consistent operations even as your workforce changes.
Our GitHub Repo has more technical information. You can also contact me or Gavin Green for more information.
Read more on MAGS at The Digital Engineer