The Future of Industrial Operations: Twins, Data, and Cyber Resilience
In a keynote address at ASKx Malmö 2026, Doug Warren of AVEVA outlined three forces reshaping industrial software — and why companies that ignore them risk being left behind.
The industrial twin — the full asset lifecycle in one place — that will be the differentiator. Not subject matter expertise alone. The twin.
— Doug Warren, ASKx Malmö 2026
The Industrial Twin as the Differentiator
For years, the competitive edge in industrial software has rested on deep subject matter expertise — product knowledge accumulated over decades, and the seasoned engineers behind it. Doug opened his address by acknowledging that reality, then immediately challenging it.
“That expertise is going to come under pressure as AI and AI agents continue to proliferate,” he told attendees. The question he posed was pointed: when AI can replicate knowledge work at scale, what remains uniquely yours?
His answer: the industrial twin — a unified, lifecycle-wide digital representation of both the physical asset and the processes it runs. And crucially, it is something very few companies can credibly offer end-to-end.
What the industrial twin is made of
Doug described the industrial twin not as a single product but as a convergence of every data layer across an asset’s life — from first pencil stroke to ongoing optimization:
What makes this more than an archive is access. The industrial twin makes all of this information available across the complete asset lifecycle — engineering, operations, and optimization — without switching between disconnected tools.
The lifecycle view matters. Doug was explicit that AVEVA’s differentiating capability is full end-to-end coverage: from engineering and design, through operations, all the way to optimization. “That’s what AVEVA does,” he said, “and it’s very much unique in the industry.” The CONNECT platform, previewed at the conference, is the thread binding those capabilities together.
Why AI makes the twin more urgent, not less
The rise of AI agents might seem to threaten the value of accumulated expertise. Doug’s argument flips that logic: AI agents are only as good as the data context they can access. A rich, connected industrial twin becomes the foundation on which AI-driven decisions are made — and that foundation is not easy to replicate. The twin is what turns AI from a generic tool into an industrial-grade one.
Industrial Data Operations and the Shift Beyond Traditional HMI/SCADA
HMI and SCADA systems have served industrial operations for decades — monitoring processes, triggering alarms, enabling human control. But Doug argued that a more expansive discipline is now emerging around them, one that reframes how organizations think about operational data entirely.
Quoting a Forbes analysis, he described industrial data operations as “the strategic orchestration of data from its creation to its end use, ensuring that various data sources drive optimized results within shortened analytics cycles.” It integrates IT, OT, and engineering technology (ET) into a unified system — and it places a unified namespace at its centre.
Where SCADA excels at real-time visibility and control, industrial data ops extends that value: it adds context, governance, and the infrastructure for AI-driven decision-making at scale. The shift is from reactive monitoring to strategic orchestration.
The technology stack evolving around it
Doug cited several developments AVEVA is actively investing in to accelerate this shift:
The acquisition of Swedish company Crosser was highlighted as a specific accelerator in this space — bringing homegrown Nordic technology into AVEVA’s data connectivity and interoperability layer. The CONNECT platform sits above all of it, functioning as the integrating layer across what would otherwise be a fragmented ecosystem.
Demonstrated in practice: AI-assisted operations
Doug walked through several live demonstrations to show what this shift looks like beyond architecture diagrams. Three stood out:
Agentic maintenance analysis. Using a large language model embedded in the operations interface, Doug prompted the system to monitor condenser performance on a turbine unit, detect fouling-related degradation, and calculate the revenue loss and cleaning payback period — in natural language, without manual configuration. The system deployed an agent, ran it autonomously over two weeks, and returned a power loss curve alongside a maintenance work order.
Knowledge Graph for tag reconciliation. A standing problem in industrial environments: SAP maintenance systems and PI historian systems name the same equipment differently. AVEVA’s Knowledge Graph uses an AI agent to reconcile those differences automatically, building a unified, queryable representation of asset data across systems.
MCP gateways on the historian. By placing a Model Context Protocol (MCP) server in front of AVEVA Historian, Doug showed how an AI assistant (in this case, Claude) could autonomously explore a glass manufacturing line — discovering 97 relevant tags, identifying three main process areas, and analysing OEE metrics — without requiring pre-configured queries or human guidance.
The common thread across all three: humans stop spending time finding and collecting data, and start spending time acting on it. Doug framed this as the core promise of industrial data ops — not replacing the operator, but freeing them for higher-value decisions.
Operations control: the product taking shape
Doug previewed a forthcoming AVEVA product specifically targeting this industrial data ops space. Built natively on CONNECT and integrating the company’s existing portfolio — InTouch, System Platform, Plant SCADA, Historian — it is designed to take users from collection and visualisation all the way through to pattern recognition, prediction, reporting, and resolution. One package, hybrid SaaS architecture, licensed through AVEVA Flex.
CRA: Why It Matters Now and What Companies Should Do Next
Speaking to a European audience, Doug used part of his address to deliver a clear message on the EU Cyber Resilience Act — a regulation that, despite its technical nature, has direct operational and commercial implications for every company that deploys industrial software products within the European Union.
Two deadlines that define the CRA roadmap
Mandatory vulnerability reporting for all products takes effect. From this date, manufacturers must actively report discovered vulnerabilities to the relevant EU regulatory body.
CE marking is required for all products placed on the EU market. Products that have not completed the conformity assessment process cannot be sold in the EU.
Doug was candid that meeting these deadlines has consumed meaningful R&D capacity through 2026. But he was equally clear that AVEVA is treating compliance not as a checkbox, but as a foundation. Penetration testing across HMI and SCADA products is underway; the Q2 2026 release of System Platform, InTouch, Plant SCADA, and AVEVA Historian will reflect the bulk of that work — even if the formal CE marking comes in 2027.
Doug’s implicit guidance: do not treat CRA as a 2027 problem. The technical and process work — product audits, pen testing, supply chain documentation, conformity assessment preparation — takes time. Companies relying on software vendors should be asking those vendors today what their CRA roadmap looks like, and ensuring their own internal processes are aligned to the November 2026 vulnerability reporting requirement.
The broader message: security is infrastructure
Doug positioned CRA compliance not as a cost of doing business in Europe, but as a signal of product maturity. In a landscape where industrial systems are increasingly connected — to cloud platforms, AI inference engines, and enterprise IT — security hygiene at the product level is inseparable from operational resilience. The CRA formalises a standard the industry should have been pursuing regardless.
For companies still treating cybersecurity as a separate project rather than an embedded discipline, the 2026 deadline is the clearest possible forcing function.
