Justifying ERP migration: the impact of agentic AI

Moving a core ERP environment to a cloud platform is already one of the biggest technology decisions an organisation can make. The cost is substantial, the disruption can be significant, and the return has to justify both. Agentic AI is adding a new dimension to that calculation.

 

If agents are going to do more than assist employees – progressing workflows, interpreting business data and executing approved actions – they need access to the systems where those processes already live. ERP platforms hold much of that context: the business logic, permissions, transactional data and controls that determine how work gets done.

 

That makes agentic AI highly relevant to ERP strategy. But relevance is not the same as inevitability. Organisations still need to decide whether access to agentic capabilities truly strengthens the case for cloud migration or whether the AI investment and the ERP transformation should be evaluated separately.

 

AI is becoming part of the cloud ERP business case

SAP shows how closely AI adoption and cloud migration can become connected. At Sapphire 2026, SAP said S/4HANA on-premises and ECC customers could gain access to selected AI scenarios if they commit to transitioning the majority of their current landscape to SAP Cloud ERP. RISE with SAP customers also gain access to the Joule Assistants portfolio.

For SAP customers already considering migration, agentic AI therefore becomes another benefit to set against the cost and disruption of transformation. There is a clear architectural rationale for this. A cloud ERP platform can give agents secure, governed access to transactional data, process context, permissions and controls within a single environment.

 

But that does not remove the scale of the investment required to get there. A heavily customised ERP estate may represent years of development, integrations and business-specific processes. Migration can require data remediation, process redesign, testing and retraining, alongside the cost of the new platform and implementation programme.

 

Agentic AI may strengthen the return on that investment, but the AI use case still needs to prove its own value.

 

Oracle illustrates why AI adoption and ERP migration should not automatically be treated as the same decision. Its Fusion Agentic Applications are built directly into Fusion Cloud Applications, where they operate within the platform’s data, workflows and security framework. But Oracle has also spent 2026 expanding ways for E-Business Suite customers to use agentic AI without first replacing EBS.

 

In March, Oracle demonstrated an agent built with its Private Agent Factory that worked with EBS procurement data. In May, it documented an architecture connecting Fusion AI agents to an existing EBS system of record without replacing it. By August, Oracle was promoting agents for EBS that could be adapted to a customer’s existing environment, data and business processes.

The wider EBS roadmap gives that development greater significance. Premier Support for E-Business Suite 12.2 is committed until at least 2037, so customers are not facing an imminent support deadline that forces them to re-platform. That gives organisations scope to assess agentic AI separately from the ERP migration decision, introducing new capabilities around an established ERP environment while continuing to evaluate whether, and when, a move to Fusion makes business sense.

 

The broader implication for CIOs extends beyond Oracle. Cloud ERP may offer advantages around native integration and access to particular AI capabilities, but the technology itself does not make wholesale ERP migration an automatic prerequisite for adopting agentic AI. The AI use case and the wider ERP transformation still need their own business cases.

 

Start with the process, then decide the architecture

This should shape how organisations build the business case.  

 

Rather than beginning with a target platform, they can begin with the process they want to improve. What information would an agent need? Which systems would it need to interact with? What should it be allowed to do, and where should human approval remain? The answers may point towards a cloud ERP platform because of the level of native integration it provides. They may also point towards an architecture that connects AI to an existing transactional core.

 

This keeps the AI investment measurable. Organisations can establish whether agents improve the process in question before using those expected benefits to support a much larger ERP transformation.

 

It also avoids treating every form of modernisation as one programme. An organisation may have a strong case for improving data access or introducing agentic capability while its existing ERP environment continues to perform its operational role. ERP replacement can then be evaluated against the problems it is expected to solve.

Support timelines still influence the decision

For SAP customers, this assessment is taking place against a much more immediate backdrop. Mainstream maintenance for Business Suite 7 core applications ends at the end of 2027, with optional extended maintenance available until the end of 2030. That timetable creates pressure to make a decision, but it does not answer the underlying question of whether the ERP environment itself needs to be replaced. A system can still be stable, heavily embedded in the business and delivering value even as the vendor changes the terms on which it will support it.

 

There is also another support option available. Organisations can move an established ERP environment to independent third-party support, allowing them to maintain it without immediately committing to a wider transformation programme. Support and transformation are separate problems. With third-party support in place, an organisation does not have to treat the vendor’s maintenance timetable as the deadline for a wider ERP upgrade. It can decide when migration is justified by its own operational requirements, investment priorities and expected return.

 

That flexibility is especially valuable as agentic AI competes for the same funding and internal capacity. Instead of committing capital to a broad ERP transformation simply to stay within the vendor’s preferred roadmap, an organisation can use that period to test AI use cases, improve the data and integrations those use cases depend on, and establish which parts of the technology estate genuinely need to change.

 

In that context, third-party support does more than extend the life of the existing system. It creates room to sequence investment differently: maintain a stable core, prove where AI delivers value, and modernise the ERP estate when the business case supports it.

 

Keep the decisions separate

Agentic AI is becoming a more important part of ERP strategy, but the developments we have seen in 2026 show that there is no single route to adopting it. SAP is linking selected AI capabilities more closely to Cloud ERP, while Oracle is also extending agentic AI into established E-Business Suite environments. That makes it difficult to argue that ERP replacement is an inherent technical requirement for agentic AI.

 

Ultimately, AI adoption, support strategy and ERP transformation should not be collapsed into one decision. Vendor support timelines may create pressure to act, but third-party support can give businesses more control over when they modernise. That, in turn, creates space to test where agentic AI delivers value, understand the architecture those use cases require and decide whether a broader ERP migration is justified.

 

The sequence matters. Establish the value of agentic AI first, then determine what needs to change in the underlying estate to support it. If that strengthens an existing case for cloud ERP migration, the investments can move together. If it does not, AI should not become the reason an organisation commits to a transformation that could not otherwise justify its cost, risk and disruption.

 

---

 

Chad Stewart is Vice President of Artificial Intelligence & Transformation at Spinnaker Support

 

Main image courtesy of iStockPhoto.com and Khanchit Khirisutchalual