The Great Unbundling: Can Agentic AI End Forced ERP Upgrades?

Joe Locandro has seen it all. For over 25 years, he navigated the C-suite across airlines, utilities, manufacturing, and construction—from Hong Kong to the Middle East, from Australia to New Zealand. He signed the checks, received the vendor pitches, and stared down yet another forced ERP upgrade.

Now, as executive vice president and global chief information officer at Rimini Street, Locandro sits on the other side of the table. What he sees isn’t pretty.

The enterprise software industry perfected a beautiful deal. Every three to five years, the big players dangle the carrot of “innovation” while wielding the stick of discontinued support. Companies spend millions on upgrades they may not need, rebuilding functionality they already have. The alternative is unthinkable: being left behind on “unsupported” legacy systems.

But Locandro has news for the ERP establishment. The game is changing. AI isn't just disrupting how we work; it’s removing the fundamental assumptions that have kept enterprises trapped in vendor roadmaps for decades.

The AI-powered innovation escape hatch

“Until recently, you didn’t have a choice,” Locandro explains, his voice carrying the weariness of someone who lived through countless upgrade cycles. “The only way to get innovation was through upgrades. You get the world’s best practice by getting the new release, and they just fuse it into the latest one.”

The psychology is elegantly simple. Vendors make “innovation” synonymous with “upgrade,” forcing customers into a perpetual cycle of paying for solutions to problems they didn’t know they had. And then create urgency by threatening the loss of support.

“Because no real alternative existed, every three to five years you had to do this massive upgrade,” Locandro continues. “That’s been a well-trodden path for CIOs like myself for the last 10-15 years.” It’s a hamster wheel disguised as digital transformation.

AI just poured cold water on this notion of forced upgrades. Specifically, the combination of generative AI and agentic AI opened a door that wasn’t supposed to exist. “You now have a combination… that can be built on top of core ERPs,” Locandro reveals. “So you can start flowing new processes and innovations without changing the core.”

Suddenly, that “legacy” ERP system, which was only one version earlier, is no longer considered tech debt. It becomes a perfectly functional system of record that serves as the foundation for transformative capabilities built upon it. No rip-and-replace, no years-long migrations, and no vendor-dictated timelines.

“Instead of having to meet a vendor roadmap or timeline to move yourself along a digital transformation path, you now have an alternative: building the new technology on top,” Locandro explains.

It also means CIOs have a choice. “The paradigm shift for companies now is opportunity cost of capital,” Locandro argues. “Do I spend more money upgrading a system over three to five years? Or do I spend it on plants, equipment, and factories to upgrade output?”

And this is not just about saving money either. “Having a third-party provider like Rimini support you for 5-10 years gives you time to make that thought process about where you want to go and where the market will be.”

Agentic AI unlocks legacy advantage

Perhaps Locandro’s most revealing insight comes when he discusses the reality facing companies already partway through vendor-mandated migrations.

SAP’s clean core requirement for cloud deployment, for example, exemplifies his argument. Although such requirements allow for core modifications, they were positioned as providing the benefits of reduced TCO, easier upgrades, and a more agile and future-proof IT landscape.

“But companies that decided to migrate two years ago, before the [ChatGPT-led] AI boom, find it very, very difficult to unravel the customizations,” Locandro argues.

Here is where Locandro gets genuinely excited as he paints a picture of AI agents that don’t just optimize existing processes but reimagine them entirely, without the need for a clean core. The agentic AI operates on top of the existing, customized system, leveraging the business logic and data already in place.

Such an approach fundamentally shifts the power dynamic. Instead of conforming to a vendor’s “clean” standard, a company can innovate on its own terms, using its existing customizations as an advantage rather than a liability. It also fundamentally alters how systems interact: instead of vendor-specific AI that works only within its silo, agentic AI orchestrates across platforms, extracting value from data wherever it lives.

Looking forward, Locandro sees this shift shaping how enterprise software works. “Your underlying ERPs will be plug-and-play in the future, as you build more functionality outside the core. You’re really shrinking the value of the ERP just to be a transactional system.”

This is already happening. Companies are building process intelligence, workflow automation, and business logic in separate layers that sit above their ERPs. The database becomes a commodity; value moves to the orchestration layer.

“And if you take the view that AI is going to get better, its ability to grab data out of different source systems is pretty easy through a data fabric, then your flexibility is exponential,” he adds.

Welcome to the great ERP unbundling

The implications for vendors are profound. If ERPs become simple systems of record, what exactly do customers pay premium prices for?

The ERP establishment isn’t waiting to find out. As Locandro observes, vendors are acquiring AI companies and building platform capabilities. However, a fundamental problem exists with vendor-specific AI: it only works within its respective vendor's ecosystem. Meanwhile, platform-agnostic AI can orchestrate across any system, creating capabilities no single vendor can match.

Whatever the near-term outcome, it’s clear from Locandro’s analysis that the forced upgrade model, which has sustained ERP vendors for decades, is colliding with a technology revolution. Although one can argue that his analysis is Rimini Street-centric, it does raise the question of what happens to upgrades when they are no longer urgent or, in some cases, even counterproductive.

AI is also altering the fundamental economics of enterprise software, which decision-makers battling AI fatigue and regulatory complexities will welcome. When intelligence can be layered on top of any system, when data can be accessed directly rather than being moved and processed, and when workflows can be reimagined rather than just automated, the old rules of ERP sales no longer apply.

The question then isn’t whether the old model of ERP sales will survive, but how quickly enterprises will realize they no longer need to be hostages.

Image credit: iStockphoto/vicnt