ERP programs force companies to make decisions today that will shape how the business operates for years. AI raises the stakes because the technology is changing faster than most operating models, governance structures, and enterprise systems can adapt to.

A company beginning an ERP transformation now may still be implementing two or three years from today, when available AI capabilities could look materially different. That makes prediction a weak basis for ERP design.

The better approach is to design around what should remain durable: be clear about the business you are trying to create while preserving flexibility in how technology ultimately enables it.

Start With the Future Business

ERP design should begin with corporate strategy and the operating model required to execute it.

If a company’s strategy calls for doubling revenue without doubling administrative overhead, the design implications can be defined before the technology roadmap is settled. The business may need more standardized processes, fewer manual handoffs, faster decisions, better cross-functional visibility and more management by exception.

Those requirements can then flow through the design:

Strategy → Operating Model → Business Capabilities → Processes & Roles → Data & Controls → Technology

The further up that chain a decision sits, the more durable it should be. Technology can change without changing the underlying business objective.

That also makes ownership important. If process decisions, data definitions and exception rules are not assigned to accountable business leaders, the ERP design will tend to preserve current departmental habits rather than enable the future operating model.

ERP design therefore does not depend on predicting which AI tools will prevail. It depends on knowing what the business must be able to do and making design choices that keep those capabilities usable as technology changes.

Process Reengineering Is More Important in AI-Ready ERP Design

Business process reengineering has always been fundamental to a well-run ERP implementation. AI does not change that principle. It makes poor process design more consequential.

Historically, weak future-state design might leave behind unnecessary approvals, workarounds, customizations or manual effort. Those same choices can now also limit what the company is able to automate later.

That puts greater weight on business process mapping that separates the underlying process from the way work happens today. The design should establish the intended outcome, the decisions and controls required to achieve it, and where human intervention is necessary. How that work is executed can then change as the technology evolves.

Consider customer order exception management. Today, an employee may identify an issue, gather the relevant information, determine the appropriate response and escalate where necessary. Over time, AI may identify the exception earlier, classify it, recommend a response or resolve certain cases automatically.

The company does not need to decide today how far that automation will eventually go. It does need a process designed clearly enough that work can shift between people and systems as the technology matures.

Build a Stable Core Without Closing Off the Future

That flexibility still depends on a sound ERP foundation.

ERP will continue to provide the transactional and control backbone for most companies, including critical operational data, core workflows and enterprise controls. Some AI capabilities will be embedded directly in ERP. Others will operate across ERP, CRM, and other enterprise platforms.

Good ERP design needs to support both paths.

Standardized processes, governed data and well-managed integration create room to evolve. Extensive customization, fragmented data and tightly coupled architecture do the opposite. The objective is not maximum flexibility. It is to avoid making decisions today that unnecessarily constrain useful technologies tomorrow.

Governance Needs a Longer-Term Lens

This also changes how ERP programs should govern design decisions.

Many choices that create future constraints look reasonable in the moment because they solve an immediate project problem. But a decision that resolves today’s implementation issue can create lasting constraints on process, data or architecture.

Steering committees should therefore evaluate significant design decisions against more than cost, scope and schedule. They should also consider whether a decision makes future integration, automation or AI use materially harder.

That does not mean optimizing every ERP decision for a hypothetical use case. It means understanding when decisions affecting process standardization, data, architecture and business logic preserve future options—and when they eliminate them.

ERP Design for AI Optionality

No ERP team can reliably predict the AI capabilities its organization will use five years from now. It does not need to.

Companies can be deliberate about the parts of the operating model that must endure: how work should flow, where decisions should sit, what data must be trusted and which controls cannot be compromised. The technology used to execute that model can then evolve.

The measure of good ERP design is not whether it anticipates every future technology. It is whether the decisions made today leave the business able to take advantage of those technologies when they become useful.