Most enterprises have been adopting AI for years, if not decades. Very few, however, have pulled off an AI-led transformation. That’s because digitization in US manufacturing has moved from the "collect everything" phase to the “build trust” phase. Most plants sit on decades of operational data trapped in historians, MES, CMMS, and SCADA systems that were never designed to talk to each other. The few manufacturers pulling ahead are treating data standardization as the foundation of their AI strategy, not an afterthought.
Furthermore, once you’re ready to graduate from pilots to enterprise-wide change management, you likely need more than a new platform. You might need an advisor, like a services company, to help you with all the required changes.
There’s so much discussion about how agentic AI and vibe coding are changing the SaaS space because these trends are impacting the development of the services themselves.
When you take a closer look at the fundamental shift these upgrades require, however, it becomes clear that the traditional principles of change management no longer apply.
Traditional principles vs. new principles
In the past, here’s how a typical change management process might go. Say you have an application and you want to add a few more pieces of information to it. You pull in the data, display the updates, and run some analysis. Typically that meant a full change management process, all the way from the data to the application. You would have to figure out all those pieces and apply changes for each module.
Compare that to what’s possible now. Once you have your data contextualization, even if your application doesn’t use the data, it’s already exposed through MCPs and ontology. This means you can simply use your favorite AI agent tool or vibe-coded application and say “Display the operator’s current certifications alongside the machine's telemetry.” A multi-week engineering ticket becomes an hours-long change.
The old model of change management treated adoption as something you handled after the technology was in. Now, the cultural shift has to happen inside the build, not after it. If you embed a center of excellence on-site from Day 1, the client's engineers can build live dashboards and work on the platform during deployment, not get trained on it once it's finished.
The whole process of point-based application development itself is changing, letting you shave your timelines down from months to days. Indeed, McKinsey is projecting that as enterprise software development moves toward multiagent coordination, organizations will achieve a 20-times leverage lift.
Forget the traditional services model
Speeding up development is only half the battle. Your company’s culture also needs to be able to keep up with the speed of that change.
Global system integrators (GSIs) will be under heavy pressure to adopt this new change management process. The ones who move early will help turn their companies into frontier AI firms. As a result, they will also be the ones who survive, while the ones who continue on the traditional path will struggle. When the old model is under pressure, it never ends well for the incumbents.
The traditional services model will shift from developing applications to focusing more on using the algorithms to build and fine-tune the AI models, as well as the data contextualization space. This structural shift will happen because application development is becoming more automated.
Consultants won't have to focus on the backend technology because their technology partner will handle that. Instead of focusing on building persona-based applications, they can focus more on data contextualization, business logic, and automating upstream and downstream processes.
The right data foundation is the key
Take a predictive maintenance application or a monitoring application. In the past, an enterprise would procure a standalone asset predictive maintenance application and a separate asset monitoring application. That’s two different applications from two different vendors, purely configured for that purpose.
Now, once you have a data orchestration, data operations, and data contextualization platform in place, you can quickly build an APM application, a monitoring application, and even an inventory management and forecasting application on top of the exact same data foundation. This simply wasn't possible before. With the right data foundation, it is. Through 2026, Gartner expects organizations will abandon 60% of AI projects not supported by AI-ready data.
A few years ago, the big consulting firms and GSIs wanted to build AI solutions themselves. They'd assemble a team, stand up custom infrastructure, and spend months on it. That math doesn't work anymore because the timelines clients expect have collapsed.
Adapt or fall behind
Every company is adopting a DataOps platform or is starting on this change management journey. There is a shift from point-to-point integration toward a contextualized, governed data layer—a unified namespace and semantic modeling, often anchored to standards like ISA-95 and CFIHOS, so that an asset or event means the same thing across every site.
A small minority have their own internal politics and departments that are still trying to make the old model work because the shift is a big disruption to their careers. That's how those companies are set up, and that's how those individuals are incentivized, so they keep putting out RFPs and not looking at the bigger picture. There will always be laggards that will catch up a decade later. That said, more buyers are writing "production AI, not POCs" into their requirements. In fact, a recent RFP required an actual "technical implementation of at least one real use case.”
If you don’t want to fall behind, start asking the right questions and adapt to the new normal.
