Amid key developments in the field of generative artificial intelligence (AI), in particular ChatGPT, technology companies of all sizes are moving to both harness its disruptive innovation and meet the evolving needs of the global business community. At the same time, corporate leaders must navigate mass excitement and concern from the public sector about the impact of generative AI across all industries.
During an hour-long July 11 virtual townhall, SAP AI experts sought to strike a balance between these strategic priorities. Presented by ASUG’s SAP S/4HANA ERP Community Alliance, the ASUG Task Force discussion (which you can watch in full here) explored SAP’s AI strategy and business automation, with two SAP speakers—Raul Porras, Enterprise Architect, SAP, and Nagi Nallamilli, Customer Officer, SAP—offering insights into how to implement business automations in SAP S/4HANA, automation trends across industries, and examples of automation use cases.
“Artificial intelligence is a hot topic. It’s something everybody’s talking about, and there have been a lot of advances front and center for the general population but more so for the business side,” Porras said, introducing the task force. “The questions are: How can we take advantage of it? How can we get ahead of it? And how do we not miss the boat?”
“Built for Business”
First reflecting on the recent surge of interest in AI technology—and on how hyperscaler availability and the compute capacity to centralize the training of models has factored into it—Porras called attention to the mainstream debut of AI technologies like ChatGPT and OpenAI. “It’s not that the technology has changed, but its accessibility accelerated adoption and enhanced innovation,” he said.
From document processing to recommendations, forecasting, generative AI, and digital assistants, the potential for business-specific AI is significant. Still, Porras and Nallamilli don’t see it as an evolution capable of replacing skilled workers. “The main reason people are considering AI is to augment what humans do,” Porras explained.
To that end, hre said, SAP AI is “built for business," being built into applications that already power critical business processes, rather than requiring customers to change their usage of those applications. SAP AI will be trained with industry insights, business process expertise, and tailored to customers’ data, to remain relevant to driving business value and accelerating business processes. SAP AI is also built on ethical and data privacy standards, ensuring decisions that come out of artificial decision-making are non-biased and respect customers’ privacy by minimizing any amount of customer data used to retrain models or provide feedback to third parties.
In discussing SAP’s overarching business AI strategy, Porras and Nallamilli named cloud ERP, human capital management, spend management, business networks, and customer relationship management as areas in which SAP foresees automation through AI-powered business processes.
Core to each of these is business data. Porras emphasized that “responsibly embedding AI capabilities in each” is intended to help companies utilize that data more effectively. To do so, SAP plans to leverage what exists and emerges from what it calls the “open ecosystem of general-purpose AI tooling,” leveraging what its partners have contributed in terms of creating, training, and running models—including those relevant to generative AI—and selecting those providers’ technologies based on usage patterns, enterprise-grade qualities, and other such factors.
To that effect, as far as its use of generative AI goes, SAP will not use ChatGPT, an application built on a machine learning Natural Language Processing model, known as a Large Language Model (LLM), developed by OpenAI. But the company will leverage that foundational LLM and its generative AI capabilities in the context of business data and processes to achieve specific outcomes in its applications, according to Porras.
To do this “responsibly,” SAP’s use of generative AI will follow the same principles of business AI, with the same development and responsible AI review processes and keeping humans in the loop to review and approve generated information. To ensure data privacy for generative AI, SAP is pursuing enterprise-ready partner agreements specific to data privacy and isolation, to ensure no customer data is used by third-party vendors to train foundational models despite its usage within SAP applications to make predictions for customers’ businesses.
"Think of AI in Layers"
Speaking broadly, Porras advised ASUG members to “think of AI in layers” within applications such as SAP S/4HANA Cloud and SAP Business Technology Platform. The digital core and backbone of the S/4HANA solution provides a framework for business processes. On top of that, intelligent technologies such as machine learning, situation handling, and analytics are fully embedded in SAP S/4HANA Cloud. Further intelligent industry capabilities like intelligent situation automation, SAP Build Process Automation, and digital assistants or chatbots are also available side-by-side via SAP BTP.
Within its focus on enriching processes and helping companies make decisions through automation, SAP is in the process of embedding AI capabilities into its existing suite of applications. Even so, this strategy is not particularly new to SAP. “Most of the applications that you own will have—or already have—artificial intelligence features and capabilities within them,” Porras said.
From predictive data analytics through statistics to making predictions with machine learning and more recently creating data models to accelerate business processes with generative AI, artificial intelligence functionality has evolved at SAP over the years. AI is comprehensive already in areas such as cash application, demand forecasting and sensing, project-cost prediction, and sales route optimization. “These features have been building up since the early days of predictive machine learning,” Porras said. Other existing use cases for SAP AI include:
- Intelligent collections and sales order auto-completion for SAP S/4HANA
- Configurable product quotation for SAP Intelligent Product Recommendations
- General ledger line-item identification for SAP Central Invoice Management
New AI capabilities that will leverage generative AI, as announced during SAP Sapphire, currently (or will) include:
- Process model generation and documentation for SAP Signavio Process Manager
- Product documentation search for SAP Digital Assistant
- Natural language queries for SAP Analytics Cloud
- Job description and interview question generation for SAP SuccessFactors
- Goods receipt processing for SAP Transportation Management
- Natural language marketing analytics, product description generation, and review summaries for SAP Digital Assistant for CX
Anurag Barua, Digital Transformation Leader, SAP, noted “a lot of demand for AI and automation across all my customers,” pointing to examples in finance such as "end-to-end automation of invoice processing, automated fraud prevention, automated financial account reconciliation and intercompany reconciliation, detection of tax compliance, automated creation of purchase orders and sales orders, and using conversational AI."
Barua reported that, on the manufacturing and supply chain side, he sees reduction of inventory carrying costs, real-time demand forecasting, and accurate delivery date prediction (improving supply chain efficiency) as examples of AI and automation already having a positive impact. "We are increasingly seeing AI/automation as a driver for digital transformation across all industries," he added.
The Ability to Adapt
As attendees asked questions about SAP’s roadmap, various topics of interest surfaced, including the importance of earning the public sector’s trust in navigating generative AI at such a protean stage of its development and focusing on change management in moving to explain and make accessible information around the technology. Additionally, one attendee reflected, process managers embedding AI and automation within existing SAP solutions has made it challenging for enterprise architects and technologists to gain visibility of all AI innovations across the SAP product suite.
More attention must be paid, the speakers agreed, to outlining strategic objectives, technical specifications, and the implications of both for businesses, employees, and the public.
Porras stressed the role that building agility within organizations can play in preparing them to consume AI. “Forecast accuracy is important, but the ability to adapt after you make a forecast is more important,” he said, by way of example. “If you keep using historical data or even sophisticated prediction algorithms, and then something changes in the environment, that forecast isn’t going to help you.”
Approximately 50 attendees—representing companies such as Apple Inc., Bristol Myers Squibb, Bumblebee Foods LLC., Constellation Brands, HP Inc., IBM, Infosys, Johnson & Johnson, Liberty Mutual Insurance, Medtronic, Messer North America, Microsoft, Pacific Gas & Electric Company, Paramount Pictures, PricewaterhouseCoopers LLC, Southern California Edison, The Nielsen Company, Toyota Motor North America, Under Armour, Vitaquest, Walgreens, xSuite, and Zep, Inc.—registered for and attended the town hall.
More information, demos, and examples on business AI are available from SAP. To watch the task force in full, click here.