Industrial Systems Engineering in the New Era of AI, Episode 2: Rethinking the Industrial Data Fabric

Welcome back to our ongoing blog series, "Industrial Systems Engineering in the Era of AI." I'm Colin Masson , ARC Advisory Group Inc. 's Director of Research for Industrial AI, and I'm continuing my conversations with industry visionary Rick Bullotta .

In our first episode and blog post, we laid the groundwork by exploring the foundational data challenges that have shaped the industrial landscape for decades. We discussed the evolution from basic connectivity to the rise of Industrial IoT, highlighting the persistent hurdles of closed systems, data duplication, and the burgeoning value of unstructured data.

Now, in our second discussion, Rick and I dive into the transformative impact of the modern AI boom—a shift catalyzed by the "November 2022 moment" with ChatGPT. This new era has reignited a focus on data quality and is forcing a fundamental rethink of how we architect our data infrastructure, which I refer to as the Industrial-grade Data Fabric.

Listen or Watch

You don't just have to read about our conversations; you can also listen in! Our discussions will be available on:

ARC ADVISORY GROUP'S SYNDICATED DIGITAL TRANSFORMATION CHANNEL (AS A PODCAST). 

ARC ADVISORY GROUP'S YOUTUBE CHANNEL (FOR A VIDEO EXPERIENCE).

Key Insights from Our Conversation

Our discussion centered on how the new wave of AI is changing not just the tools we use, but the very language and approach to solving industrial data problems. The conversation has shifted from legacy terms like "data historians" to the more holistic concept of the Industrial-grade Data Fabric.

A New Spirit of Openness

What makes this wave of AI feel different from previous hype cycles is the cultural shift it has inspired. It's not just about the technology of Generative AI itself, but the renewed momentum it has created around solving long-standing problems of interoperability.

Rick Bullotta: "The big change is that the conversation has moved from just connecting systems to understanding what's possible when you have a truly intelligent data foundation. We're not just looking at data; we're looking at what AI can do with that data to fundamentally change how we work."

Colin Masson: "The November 2022 moment had a tremendous impact, but it's not just about a specific large language model. It's everything that moment triggered—a much greater willingness across the tech industry to think about interoperability, embrace open systems, and solve large-scale problems collaboratively."

Moving Beyond the Data Lake: Assembling the Fabric

A central theme was the evolving consensus that you don't need to move all your data to a central cloud repository. The cost and complexity of moving petabytes of data—often without context—are prohibitive. In practice, this means companies are not buying a single "fabric" but are actively assembling one.

Rick Bullotta: "The idea that you must ingest everything into a massive data lake in the sky is being challenged. We're realizing that data has a half-life, and its value is often highest at the source. The question is no longer 'How do we move all the data?' but 'Where does the data need to live to be most effective?'"

Colin Masson: "What I see in practice is that larger enterprises are assembling what I view as an Industrial-grade Data Fabric. On the enterprise side, the choices are more predictable, but on the industrial side, it's far more fragmented. People are having to assemble it piece by piece, asking questions like, 'I need a Unified Namespace for my factory data. Who am I going to use for that?'"

GenUI: A Cure for Configuration Headaches?

Another exciting development we explored is "GenUI" (Generative User Interface). For years, the immense complexity and configuration demands of industrial software, such as Manufacturing Execution Systems (MES), have been a major barrier to scalable adoption. Gen UI promises to radically simplify these user interactions.

Rick Bullotta: "Gen UI has the potential to solve the complexity problem that has plagued industrial software for years. Instead of spending months on configuration with highly specialized experts, you could have a system that understands natural language and configures itself based on user intent. That's a game-changer."

Colin Masson: "In our sector, we deal with very complex systems that require lots of configuration because processes are not highly standardized. In many ways, that's been the biggest barrier to having MES at scale. Gen UI can take all of that complexity away, if it's done right."

The Pragmatic View on Data Quality

Finally, we tackled the "religious war" over data quality. Does data need to be perfect before applying AI, or can AI help manage imperfect data? We agreed that striving for perfection often leads to "analysis paralysis."

Rick Bullotta: "The debate over perfect data versus 'good enough' data is critical. Striving for perfection can lead to analysis paralysis. Modern AI is resilient; it can often filter out noise and even help identify and clean up data quality issues."

Colin Masson: "There is an approach where you just look at the data quality you need for each use case, and use each one to start building out that future data fabric over time. That seems more practical than waiting two or three years on a massive data cleansing project before making any progress."

From Assets to Processes: A Crucial Mind Shift

Our conversation also underscored a critical mind shift from a purely asset-centric view to a more holistic, process-centric one. While assets are tangible and have sensors, real value lies in understanding the entire process.

Rick Bullotta: "We have to start thinking about the intersection between asset data and process data. People build great models for a pump, but processes are so diverse. I just don't see enough effort going into process data analytics, and that's where the two worlds must meet."

Colin Masson: "There's so much data that is really about the 'how, why, when,' and the quality of what's flowing through the process. It's not about the asset; the asset is just a means of making the product. We need cost data, energy data, and schedules to know if we're truly performing."

This second conversation reinforced that we are at a pivotal moment. The new era of AI is not just adding another layer of technology; it's forcing us to re-evaluate and re-engineer the very foundations of our industrial data systems.

In our next episode, we'll get even more specific, exploring the components of a modern solution architecture, the future of applications like MES in an agent-driven world, and the practical steps for building out your own Industrial-grade Data Fabric.

Diving Deeper: Essential Reading

As we venture further into topics like building robust data infrastructures and modernizing architectures to effectively infuse AI, I highly recommend readers explore some of my existing research:

• My blog series on "Assembling Industrial-grade Data Fabrics."
• My series on "The Rise of A2A: Completing the Industrial AI Protocol Stack with OPC UA and MCP."
• The article, "Core Capabilities of the Industrial-grade Data Fabric: Powering AI Infusion and Modernization", which is the sixth post in the Data Fabric series and delves into the essential solution services these fabrics enable.

These pieces offer a solid foundation for the themes Rick and I will continue unpacking in this series.

Engage with ARC Advisory Group

For ARC Advisory Group recommendations for navigating the AI Wars, closing the digital divide by embracing Industrial AI, assembling your Industrial-grade Data Fabric, the modern Industrial AI technology stack, and governing and guiding major decisions about enterprise, cloud, industrial edge, and AI software, please contact Colin Masson at cmasson@arcweb.com or set up a meeting with me, or my fellow Analysts at ARC Advisory Group.