Companies skipping SAP’s Joule AI in challenging S/4HANA transition

The Swiss food giant Nestlé is demonstrating how it can be done better. The company is currently migrating to SAP S/4HANA and integrating SAP’s AI assistant Joule directly into the system. But many companies are struggling with the transition to an automated SAP future, as a recent market analysis by management consultancy Horváth shows.

Six out of 10 companies that are also in transitioning to S/4HANA and could implement Joule without significant additional resource expenditure consider themselves “not agile, efficient, and flexible enough” to manage couple SAP’s flagship cloud ERP with its AI copilot offering.

Horváth’s results are based on a survey of 200 companies worldwide with an annual turnover of €200 million that are currently converting their business processes to SAP S/4HANA — a daunting journey that many companies are struggling with and others are questioning the ultimate ROI.

S/4HANA as a starting point for AI tools

According to Stefan Maus, SAP expert and partner at Horváth, the migration to S/4HANA is actually a good starting point for directly planning and implementing the integration of AI tools, such as Joule and AI agents from SAP. The ERP giant is betting big on both technologies, with plans to make Joule ubiquitous across its portfolio, and having announced its all-in intentions on agentic AI at this year’s Sapphire conference.

“However, this aspect is usually neglected during the transition, or organizations fail to integrate the topic holistically and thoroughly,” Maus said.

Horvath points out that Joule has demonstrated significant efficiency gains in pilot projects. In purchasing, for example, it can automatically obtain quotes from suppliers, compare them, and generate a purchase order proposal.

Even without AI, it’s a challenge

But while AI is at the top of CEOs’ agendas for IT, many companies are not implementing AI during their S/4HANA migrations because project teams generally need more time and budget to complete the ERP transition work itself.

According to the survey, even without extensive AI add-ons, system migration often proves more challenging than initially anticipated. Nearly half of all companies (46%) stated that they would subsequently allocate more time and a larger budget. This, they explained, is the only way to ensure that all requirements associated with an S/4HANA transformation are met and, above all, that sufficient time is available for adjustments and optimization.

AI optimization is still possible after the switchover

While the targeted use of Joule and AI solutions is particularly effective in greenfield scenarios, where processes are set up and optimized before the system changeover, Horváth’s Maus also believes it makes sense during the optimization phase to individually examine and selectively utilize the functions of Joule — which is fully integrated into the system and included as standard in various license models — for standardization and automation.

Philipp Herzig, CTO and CAIO at SAP, generally agrees. “Of course, many companies are currently heavily involved with the transformation. It’s a busy time — often a mammoth task,” he explained in an interview with Computerwoche. Nevertheless, experience shows that it makes sense to start implementing Joule early, ideally in parallel with the migration to RISE or S/4HANA.

“Many of the embedded AI scenarios in S/4HANA — whether in the private or public cloud — are only fully available once the transformation is complete. However, companies can take action beforehand: Most already use other SAP systems in the cloud, and Joule can often be used there already,” Herzig said.

Minimal costs, effort, and risk

According to Horváth, what makes Joule’s use particularly attractive are the low or nonexistent acquisition costs. For example, the AI copilot is already fully included in the RISE license model in the public cloud, and at least with basic functions in the private cloud.

Furthermore, there is no implementation effort required, as Joule is already fully integrated into the S/4HANA environment. No interfaces need to be created either. “There is virtually no risk — except for getting bogged down in details,” Maus stated. He recommends initially focusing on mass processing or processes with standard use cases where its implementation promises the greatest immediate benefit.

“It’s better to select specific business processes and automate them properly and completely than to do everything a little bit,” he explained.

Gradual introduction instead of a big bang

“The best strategy isn’t a big bang, but a phased approach,” confirmed SAP CAIO Herzig. “We at SAP did the same: First, we integrated Joule into SuccessFactors, then across our entire portal, and so on. In the next step, we integrated our policies, later Ariba, Concur, and finally S/4HANA Private Cloud. With each additional system that is integrated, Joule becomes more powerful and useful for more employees.”

Initially, perhaps only a few hundred people will use the features daily, later thousands — simply because the added value grows with every step.

The same applies to the Business Data Cloud (BDC), Herzig added. “It too can be introduced gradually — initially in one area, later more and more broadly. If you migrate old systems or data repositories step by step, BDC automatically becomes more powerful and useful,” he said.

What’s important is expandability, Herzig added. “You can already do this during the transformation. Create additional connections to legacy systems if they contain particularly important use cases. If the functionality later becomes standard practice from SAP, this in-house solution can be easily replaced — without significant effort and with a clear TCO advantage,” he said.

No matter how you approach it, in the long run, SAP users will likely have to deal with Joule and AI agents. “In the next five years, Joule will be an integral part of business processes in companies that use SAP software,” Horváth manager Maus is convinced.