August 21, 2026 • 10-12 Min Read

No CEO says, “Let’s spend millions so people can create new spreadsheets after go-live.” No CFO says, “I hope the close takes longer.” No CIO says, “What we really need is another integration mystery no one can explain.”

And yet, that is exactly what happens in too many ERP programs. Somewhere between the boardroom promise and the go-live reality, things get messy. The system technically works, but the business struggles to run. Reports do not tie out. The close drags. Inventory behaves differently than expected. Users quietly create workarounds. Integrations fail at the worst possible time (apparently integrations have a sense of humor). Finance starts questioning the numbers. Operations starts questioning the process. IT starts questioning its life choices.

Then comes the familiar phrase: “We need stabilization.”

Of course we do. But here is the uncomfortable truth: stabilization is often just the invoice for ambiguity that was never resolved upstream.

ERP programs rarely fail because the software cannot process a journal, purchase order, invoice, shipment, customer, vendor, item, or asset. ERP programs fail because the implementation begins with an incomplete understanding of how the business actually operates.
That is the real issue.

The Business as Described Is Not Always the Business as Operated

Traditional ERP discovery relies on workshops, interviews, process walk-throughs, requirements templates, system inventories, spreadsheets, and the collective memory of people who are already busy running the company.

Those inputs are necessary. They are also dangerous if we pretend they are complete.

People describe the process they believe exists. Or the process they wish existed. Or the process they follow on a good day, in one region, when the data is clean, the customer behaves, and everyone remembers the workaround.

That is not always reality.

In my experience, the real business usually lives in the gaps. It lives in the spreadsheet someone built five years ago because the report was wrong. It lives in the access database no one wants to admit still exists. It lives in the manual reconciliation that “only takes a few hours,” which somehow requires three people, two late nights, and a minor act of divine intervention. It lives in the custom report that no one owns but everyone depends on. It lives in the exceptions. And ERP is where exceptions go to become expensive.

The problem is that the truth often shows up only after the program is already moving. And sometimes the truth arrives after go-live. By then, assumptions have become configuration. And go-live has become a hostage situation. The later the gap between the business as described and the business as operated is discovered, the more expensive it becomes.

That is why ERP needs a new starting point: a contextualized, evidence-based view of the operating landscape before blueprint and design.

That is where AI changes the game.

AI Does Not Replace ERP Judgment. It Removes Avoidable Guesswork.

Let’s be clear. AI is not going to magically implement ERP. The stronger and more practical opportunity is to use AI to rapidly scan, organize, and contextualize the evidence of how the enterprise actually runs. Then experienced ERP, finance, supply chain, data, controls, change, and technology leaders can interpret that evidence and make better decisions earlier.

AI can inspect configurations, transaction patterns, data, integrations, reports, tickets, workflows, controls, customizations, and recurring defects. It can spot inconsistencies, hidden dependencies, duplicate logic, shadow processes, and places where the documented process does not match operational reality.

AI’s evidence base changes the quality of the conversation.

Instead of asking, “What do people think is happening?” the program can ask, “What does the evidence show is happening?”

That is a very different starting point.

AI also addresses one of the fundamental problems in ERP: translation.

The business explains what it needs. Functional teams translate that into requirements. Architects translate requirements into design. Configuration teams translate design into build. Test teams translate build into scripts. Change teams translate scripts into training. Users translate training into behavior.

That is a lot of translation.

AI-enabled discovery now gives teams something they have rarely had before: a common body of evidence against which assumptions, requirements, designs, and decisions can be tested.

The First AI Implementation Should be the ERP Implementation

There is another reason this matters now. ERP is no longer just a back-office transaction platform. It is becoming part of the operating context for enterprise AI.

AI ambitions depend on the quality of the ERP foundation. If the ERP program produces fragmented data, unclear ownership, weak controls, inconsistent master data, poor process discipline, and manual reporting workarounds, the company’s AI agenda starts on sand.

AI will not fix a broken operating model. It will expose it. In some cases, it will amplify it.

That is why I believe the ERP implementation itself should be treated as the first AI implementation. ERP design determines whether future AI has trusted data, governed processes, clear business context, and meaningful control points.

Traditional ERP digitizes the known process. AI-enabled ERP challenges whether the known process is true.

Before major design decisions are locked in, companies need a more complete view of the operating landscape: the systems and workflows, the data and integrations, the controls, recurring defects, spreadsheet dependencies, critical reports, process ownership, and the information the business will eventually expect to use for analytics, automation, and AI.

This is not just a technology scan.

It is an enterprise operating scan.

Better Design Decisions Are the Real Prize

ERP programs are shaped by a relatively small number of decisions with enormous consequences: how the enterprise structures its data, entities, products, customers, suppliers, approvals, reporting, security, integrations, and process ownership.

Those decisions become the architecture of how the business runs.

AI-enabled scanning helps leadership see the blast radius of those decisions before they are finalized. That is where AI creates leverage.

It can show how current-state complexity flows into design, how design choices affect testing and controls, how data quality affects reporting and automation, and where decisions being made today will either enable or constrain future AI.

ERP methodology still matters. Structure matters. Sequencing matters. Governance matters. Deliverables matter.

But methodology alone does not guarantee understanding.

A methodology can tell you what meeting to hold. It cannot tell you whether the business told the truth in the meeting.

The next generation of ERP transformation should therefore behave less like a linear implementation and more like a learning system—one that continuously connects discovery, design, testing, deployment, stabilization, and value realization.

AI can help analyze the evidence, trace decisions, expand test coverage, identify emerging risk, and surface value opportunities throughout the lifecycle.

Stabilization Should Not Mean Delayed Discovery

Every ERP go-live needs stabilization. That is normal. But there is a massive difference between expected stabilization and avoidable chaos.

Expected stabilization means the team knows the likely pressure points. The business understands the operating changes. Critical integrations are watched. Data issues are visible. Defects are triaged by root cause. Leadership understands the business impact.

Avoidable chaos is when the organization discovers after go-live what it should have understood before blueprint. That is not stabilization. That is delayed discovery with a bigger invoice.

AI-enabled discovery will not eliminate stabilization, but it can change the nature of it—from a rescue motion into an activation motion.

Instead of asking, “What broke?” the team can ask: What known risk materialized? What does the evidence tell us? Who owns it? What is the fix path? What does it mean for business value?

That is a much better conversation.

It is also a much better executive conversation.

What This Means for the CIO, CFO, and CEO

For the CIO, the mandate is not simply to implement a platform. It is to help the enterprise understand the operating complexity that the platform will either simplify or preserve.

That means partnering earlier with the CFO, COO, CHRO, business leaders, audit, data leaders, and the transformation office. It also means asking harder questions before design begins: What complexity are we actually removing? What are we standardizing, and why? Where are exceptions creating real value—and where are we keeping them because no one wants to have a hard conversation? What future operating model are we actually building?

For the CFO, this changes the economic lens.

The question cannot only be, “What will the ERP cost?”

The better question is, “What will this ERP enable?”

What manual work will disappear? What reporting friction will be eliminated? What controls will become stronger? What data will finance trust? What decisions will improve? And what will it cost later if unnecessary complexity is preserved now?

ERP design shapes close, consolidation, forecasting, working capital, margin visibility, compliance, auditability, and management reporting. In plain English, it shapes whether finance can actually help run the business instead of spending its life reconciling the business.

For the CEO, the message is simpler still.

ERP is not a back-office modernization project. It is an operating-model decision. It determines whether the enterprise can scale, integrate acquisitions, enter markets, improve productivity, automate intelligently, manage risk, respond to customers, and use AI to make better decisions.

If leadership treats ERP like a technology refresh, the company will get a technology refresh. If leadership treats ERP as the foundation for intelligent operations, it has a chance to create something much more valuable.

Those are very different outcomes.

The New ERP Starting Line

AI-enabled discovery gives companies the ability to see their operating complexity before they encode it into a new system.

That is the breakthrough.

Before blueprint, before design, before configuration, before the implementation engine starts moving at full speed, leaders need a clear view of the business they are actually transforming.

Because the goal is not to implement ERP. The goal is to create an enterprise that operates with less friction, makes better decisions, adapts faster, and can use AI with confidence.

That begins before blueprint. It begins with the truth. And if we are being honest, the truth could have saved a lot of go-live weekends.