GenAI dominates the narrative in ERP, but what is it good for?

With vendors obsessed with adding generative AI to everything, does it really have a place in ERP software? It's very early days, say analysts at Forrester.

The leading ERP vendors have stapled GenAI to their product portfolios in the last year. German software giant SAP has promised to sort and contextualize data from multiple systems. At SAP Sapphire 2024, the vendor announced Joule's inclusion in ERP system SAP S/4HANA Cloud solutions, extension development environment SAP Business Technology Platform, low-code tool SAP Build, and SAP Integration Suite. SAP Ariba and SAP Analytics Cloud will get the tool next year.

In March, Oracle launched 50 use cases for GenAI on its Fusion Applications suite, promising to provide financial report "narratives," help filter job candidates, and provide product descriptions. Infor and Unit4 are also promising GenAI features.

"Most vendors are touting investments, venture funds, partnerships, and early functionality or roadmaps in their customer communications and earnings calls. In stark contrast, customers are still very early on their ERP AI journey," Forrester said in a recent report.

While ERP users are excited about AI's potential to "drive efficiency and improve results," they struggle to understand where it fits in. They were also wary of falling for "false promises and hype that won't deliver results," it said.

Forrester sees early applications of GenAI in summarizing financial performance.

• Asda IT staff shuffled off to TCS amid messy tech divorce from Walmart
• SAP customers may struggle to escape ECC before support shutters if they don't start now
• More than a Joule required to power up enthusiasm for SAP's AI strategy
• SAP cloud swells its topline, but profits slide

"GenAI will revolutionize those types of tasks by helping generate content faster and, ultimately, more accurately (noting that training and accuracy is a curve). GenAI is already being used for narrative reporting (for investor calls or other stakeholders), for customer collections (dunning) emails, and for account summarization," the report said.

GenAI also promises to bring users closer to application development. "We are already seeing coding 'wizards' that let business users do point-and-click while also viewing code, which they can later show to developers if necessary," according to the research.

The analysis, led by Forrester veep and principal analyst Liz Herbert, said Oracle's Netsuite used GenAI for code development by mining community code for input.

Although very much en vogue, even the best GenAI models sometimes hallucinate – i.e. malfunction – and unpredictably churn out nonsense.

As such, the report warned that users should be aware of the risks.

"AI is not yet ready to be off on its own," the report said. "It does not need to go fully automated just yet. Humans should still be kept in the loop to review AI recommendations – whether those are categorization, text creation, or others."

It also said users should not be in charge of governance. "Many vendors have a click-through agreement when you want to use AI, which you'll likely want to review – not have business users clicking through willy-nilly without understanding the consequences and without legal, risk, and IT being aware." ®