.svg)
*All opinions expressed in this piece are those of Jim Jacob, and do not necessarily reflect the views of any organization.
Quickly 'bolting on' AI isn't possible for organizations operating in the heavy industrial sector. While AI headlines continue to contribute to the hype and false promises of overnight transformations and rapid results, industrial organizations are moving at a different pace than the digital-native enterprises dominating most of the AI conversation. Now, they must approach AI more deliberately, with product reliability, customer trust, and repeatable results taking precedence over speed.
For an expert's take, we spoke with Jim Jacob, Executive Director of Digitalization, Advanced Analytics, and Artificial Intelligence at Cummins Inc. Jacob has spent decades shaping enterprise-level AI in industrial settings, including architecting the first enterprise Digital Twin platform at 3M. For him, the question isn't if AI can transform business but how to do it reliably, repeatedly, and economically. "Living in an industrial, credible, and reliable company, I find it difficult to bridge the AI adoption gap at the speed implied by the headlines. It's going to take time for us to bring these new AI technologies into our systems, test them, and offer them to our customers as a service."
First, Jacob aimed at a common assumption that companies can easily monetize AI by charging customers a premium. Comparing it to the environmental responsibility movement a decade ago, when the market saw customers quickly absorb new standards into their baseline expectations.
An inconvenient truth: When discussing how AI features and functionalities are being pitched to customers and consumers, Jacob recalled that customers pushed back against environmental premiums, saying, "It's your duty to be socially and environmentally responsible. I believe the product you give me is going to cover all that. So why are you asking me for a premium?" Instead, Jacob noted, the real ROI from AI for industrial enterprises is experienced through internal efficiency gains rather than a new line item. "I think the monetization component is in the company, where efficiency and productivity can grow because of AI capability." As for charging customers more, Jacob added, "I doubt very much that that's the case today."
The market dynamic of not charging more for something that is perceived as standard, informs his skepticism of the vendor market. Jacob describes the current vendor market as populated by players making baffling or simply impossible promises for where AI is today. He recounted a recent meeting where a vendor, after hearing about a thorny supply chain problem, promised a solution in "ten days." To navigate this market, Jacob uses a simple but effective tactic.
Calling their bluff: Now Jacob’s first question to any vendor who approaches the team is direct: "Where did you do that before, and I want to speak to the subject matter expert who can tell me that you guys solved this for them." The response is typically "Crickets," Jacob said. In his opinion, you have to be skeptical because this is often how bad projects can get approved across enterprises, particularly when leaders are not deeply familiar with the underlying challenges.
Pirate Land: From Jacob's perspective, with vendors promising overnight transformations and CIO-level pitches that skip over the operational realities, companies risk buying into solutions that never move beyond proof of concept. The danger, he warned, is that executives sign checks for flashy AI projects without securing proof of value. The result isn’t just wasted investment, it's the fuel that further erodes credibility inside the enterprise, slows adoption, and creates skepticism that stalls more promising AI initiatives down the road. "It's unfortunate, but it's a pirate land right now," Jacob said.
As a response to this chaotic market, Jacob has pioneered what he calls a "digital core." Simply put, a digital core is an information fabric built on reusable ontologies that can persist across an organization. His approach is primarily designed to overcome a challenge many companies face: siloed problem-solving.
One size fits most: By creating a platform that transforms disparate data into a connected engine, the digital core system provides both immediate financial insight—allowing his team to calculate the enterprise-wide cost of a tariff "with one click"—and the operational agility to redirect distribution centers effortlessly in the face of oncoming hurricanes for example.
The reusable playbook: In Jacob's experience, he sees too many enterprises chasing one-off AI experiments that might solve a single problem but can’t be applied anywhere else across the enterprise. "The big miss for many enterprises is that they solve problems in a silo without considering attempting to solve the problem in an interconnected way," he said. Jacob calls these “seat license solutions" that are expensive, brittle, and quickly obsolete. At Cummins, his team takes a different approach in building reusable connectors that can be applied across multiple functions, from sourcing to logistics to critical parts management. This architecture framework helps to ensure each new problem solved strengthens the broader digital core, compounding value rather than duplicating effort.
Temporary challenges: Jacob was quick to acknowledge that even Cummins’ carefully architected digital core may not be the long-term end state. In fact, he predicts today’s painstaking work of modeling, connecting, and building will soon be replaced by plug-and-play solutions and reusable models, much like how developers share open-source code on GitHub. "All of this modeling, building, and connecting is going to be old-fashioned in a year or two," Jacob said. The challenge for executives is navigating this messy, transitory period: building enough reusable structure to deliver ROI now, while preparing for a near future where those foundations are automated away.
Jacob’s advice for executives and his AI-focused peers is to concentrate on building a genuine foundation that can contribute towards lasting value by championing a more agile and modular approach. Even though this advice represents a clear departure from the kind of expensive, monolithic data projects that were common in the past, he believes in creating something that enables companies to grow their capabilities without the pressure of trying to "boil the ocean."
Data fabrics to information fabrics: Jacob's approach and methodology are intentionally simple and repeatable. Enterprises need to start with a single problem, solve it by pulling the right information set through an ontology, and then reuse that solution across the enterprise. "Over time," Jacob said, "you will make a snowball big enough where you have an information reference hub that brings all the information to you whenever you need it."
For Jacob, succeeding with AI is focused on the age-old phrase of "building something that lasts". By solving real problems now, enterprises see immediate value and can demonstrate a clear return on their AI investments. At the same time, each solution adds to a reusable information fabric that makes future projects easier and more impactful. In other words, the work done today doesn’t just fix today’s challenges, it sets the stage for long-term success as AI continues to evolve.
