ERP decisions have always been long-term commitments. The systems underpin finance, supply chain and operational processes, hold critical data and often stay in place for ten years or more.
AI is adding uncertainty to those decisions. Vendors are building AI into their platforms, specialist providers are developing new approaches to automation, and it remains hard to say how these developments will reshape enterprise architecture over the next five years.
An ERP and AI strategy therefore has to work with what is not yet known. We think the most useful distinction is between investments that deliver demonstrable value today and those whose justification depends heavily on one particular vision of the future.
What is known so far
Informed views differ. In May 2026, McKinsey argued that the question is no longer whether AI will change ERP, but how radically and how quickly. It describes a range, from traditional ERP fading as AI agents recreate its capabilities, to a stable backbone that remains essential for reliability, auditability and compliance. Its own view sits in between: a hybrid approach, in which the transition “won’t happen overnight” and ERP remains in place while AI is layered around it.
The evidence on returns is more mixed. In June 2025, Gartner predicted that over 40% of agentic AI projects will be cancelled by the end of 2027, citing escalating costs, unclear business value or inadequate risk controls. The same release expects a third of enterprise software applications to include agentic AI by 2028, up from less than 1% in 2024.
McKinsey’s research from January 2026 adds that around 40% of organisations report an EBIT impact from AI, but most attribute less than 5% of EBIT to it.
Adoption is moving quickly, the end state is uncertain, and returns so far are uneven. That is a weak basis for any single bet.
Three plausible futures
We find it more useful to describe plausible futures than to predict a winner. These are our own framing, and they are not mutually exclusive. Different industries, and different parts of the same business, may move at different speeds.
There is not enough evidence to say which will dominate by 2030. The sensible aim is to avoid ERP investments whose value depends on only one of them.
What AI could mean for design and support costs
Alongside the question of architecture sits a question of cost. AI may change what it takes to design, build and support an ERP landscape, and that is worth allowing for in business cases being written now.
On design and delivery, McKinsey’s research suggests AI agents could reduce the effort needed to implement ERP systems by at least 50%. Its examples include design and build shrinking from six to nine months to two to three months, and roughly 80% less testing effort. McKinsey presents these as the potential of AI agents, based on its own research, and expects change management to be the major constraint.
That does not mean every programme will cost half as much. It does mean estimates deserve questions. How much of the proposed effort reflects work that is genuinely required? How is AI-assisted tooling reflected in the plan? And how will any savings be shared?
The same applies to support. McKinsey reports that customer expectations for year-over-year efficiency improvements from application maintenance providers have roughly doubled, and that customers will expect those efficiencies to be passed on to them.
For organisations paying for support, whether from the vendor or another provider, that raises a practical question: how much of what you pay for do you actually use? AI-assisted analysis of system usage, custom code and incident history may help establish what level of support is genuinely needed, and leave you better informed when comparing your options. We see this as a promising area to test, not a proven saving.
Support still matters. McKinsey notes that companies are unlikely to want to run critical infrastructure without vendor support, and regulated businesses will continue to depend on updates for compliance and security. The aim is support that is right-sized and better understood, not simply smaller.
Evaluating vendor AI strategies
Vendors have a legitimate interest in explaining how customers can benefit from new capabilities, and their roadmaps show where innovation is heading. They are useful information, although they are not a neutral forecast.
SAP offers a recent example. At Sapphire in May 2026, SAP offered some Joule AI assistants and agents to ECC and on-premise S/4HANA customers who are committed to moving to the cloud, as reported by The Register and CIO. The Register reported SAP’s CEO describing this as an interim solution while customers modernise. The reported conditions differ between accounts, so they, along with cost and timing, need to be confirmed with SAP before a plan is built around them. For organisations already considering a move, this may strengthen the business case.
More broadly, the value of any AI capability depends on the environment it is deployed in, including the quality of data, licensing arrangements, integration requirements and the capacity to implement. A capability that works well in one organisation may need considerable additional work in another.
It is also worth asking whether similar benefits could come from specialist applications that operate alongside the ERP. Gartner has warned about “agent washing”, where existing products are rebranded as agentic without substantial new capability. That is a further reason to ask what is available today, and what evidence there is of results in comparable organisations.
Foundations that stay valuable
Some investments are likely to pay off however AI develops.
All three depend on a clear view of the existing estate: which processes and systems are stable, where complexity has built up and what genuinely needs to change. They also depend on the skills and governance to evaluate new technology and keep automation within business controls. None of them requires a prediction about which AI platform will prevail.
A practical test for ERP decisions
For any significant ERP decision, whether to transform, phase, optimise or retain, we suggest four questions.
Retaining a stable platform can be a sound choice, provided there is a credible plan for support, security, regulatory compliance and skills. It works best where the extra time is used to improve data, simplify processes and strengthen integration.
For SAP ECC customers, timing adds a further question. Mainstream maintenance ends on 31 December 2027, with optional extended maintenance to the end of 2030. Some businesses have compelling reasons to move to S/4HANA, while others will benefit from a phased approach or a defined period of optimisation. Our Insight on SAP ECC support beyond 2027 looks at these options in more detail.
A strategy built around business value
Helixr is independent and system-agnostic. We work with SAP and other enterprise technology providers, and we help organisations understand which capabilities belong in the core, where specialist applications may offer an advantage and how the components should work together, without assuming that any one platform is the answer.
Nobody can predict exactly how enterprise systems will operate in 2030. Organisations that understand their landscape, invest in data and integration, and decide against clear business objectives will be better placed to adopt new capability as it matures.
If you are weighing up an ERP decision and would like an independent view of how AI affects it, we would be happy to talk.
Sources
McKinsey, “The end of ERP as we know it? Five ways AI is disrupting ERP”, 11 May 2026.
McKinsey, “Bridging the great AI agent and ERP divide to unlock value at scale”, 9 January 2026.
Gartner, “Gartner Predicts Over 40% of Agentic AI Projects Will Be Canceled by End of 2027”, 25 June 2025.
The Register, “SAP U-turn brings AI features to ECC and on-prem S/4HANA”, 13 May 2026, reporting remarks by SAP executives at Sapphire.
CIO, “SAP’s AI offer to legacy customers comes with a catch”, 13 May 2026, reporting SAP’s Sapphire briefing.
Through planning and due diligence to execution and stabilisation
Understanding the levers for successful integration
Expertly navigate the challenges and risks
Unpick the opportunities to drive the bottom line

Disentangling a well-established site from a complex corporate infrastructure-with stringent timelines, local compliance challenges, and a rigid transitional services agreement (TSA) in place.
Expertly navigate the challenges and risks.
Combining processes, systems and people to deliver maximum results
Working with what you've got to make things even better
Making your data work for you to deliver
real insight

How we streamlined and future-proofed, a soon-to-be obsolete labelling solution and a rapid client expansion across multiple new sites.
Enhancing decision making and strategic
alignment to drive performance
Using the latest technology to drive efficiencies, innovation and operational excellence
Meeting local requirements while
delivering smooth cross-border
operations
Simplifying supply chain finance
Improving compliance accuracy and
automation using tax technology