SAP’s Global AI Head on SAP Automation in Action
Meet the Authors
Key Takeaways
⇨ AI outcomes can be delivered strategically with three levels: Relevant, Reliable and Responsible.
⇨ SAP is rapidly moving into meaningful adoption of Generative AI to deliver value.
⇨ Generative AI makes development more accessible and efficient for individual programmers and enterprise teams.
Last year was defined by AI on the software front in enterprise technology. SAP was no exception in this regard, appointing Walter Sun as its global head of AI in the fall. For 2024, the message from Insiders is clear – a need not just for product announcements, but tangible examples of results from machine learning and Generative AI. What, users are asking, can they expect both now and in the future from AI in their SAP software? What are realistic outcomes with the tech and how does SAP see its long-term vision for artificial intelligence?
Walter Sun provides insight in this exclusive interview with SAPinsider, touching on case studies involving accenture, AI’s benefits for manufacturing and payment processing, and the use of SAP’s brand-new Joule co-pilot.
Giacomo Lee (GL): AI is not new for SAP. But how did 2023 prove different for SAP in bringing its AI wares to market?
Walter Sun (WS): SAP is focused on delivering AI for business, and our intent is to deliver the first, true system of intelligence to fundamentally change the way companies operate, enabling them to create greater value, more efficiently. In this approach, AI is infused throughout every application and at every interface that matters, with contextualized insights just a prompt away.
SAP is uniquely positioned to deliver real results with Generative AI and we are already bringing this vision to life for our cloud user base of nearly 300 million. We own the broadest suite of business applications on the planet. We don’t just offer a single solution, like a CRM. We can deliver AI outcomes across our entire portfolio, powered by the cloud and the SAP Business Technology Platform, BTP.
GL: How is that delivered strategically?
WS: From a strategic level, we do this with AI that is Relevant because it works within the context of customers’ real-world business processes and embeds into the SAP systems they use every day.
Secondly, it’s Reliable AI because it understands not just the business data, but also the business process and business context of that data. It’s one thing to have data – it’s another thing to understand what that data means within the business process.
Lastly, it’s Responsible, meaning it’s delivered with the highest levels of concern for security, privacy, compliance, and ethics. Protecting customer data is of utmost importance. For Generative AI, we provide guarantees that no LLM will use our customers’ data to learn or improve on their foundation models. And we partner with leading AI vendors to bring the right AI models to the right scenario.
GL: Is there a key change here when talking about the “generative” aspect to AI, as opposed to the AI SAP users may have previously known?
WS: AI systems are used to analyze existing data and predict trends or outcomes. They can be trained to follow specific rules, do a particular job, and do it well, but they don’t create anything new. With the advent of Generative AI, the world finds itself at the beginning of a major cycle of technological innovation. Here, we are rapidly moving past the hype phase and lots of various pilots, into meaningful adoption of Generative AI to deliver value.
This is exactly what SAP is delivering. For example, this quarter we are coming out with our regulatory change manager embedded with our new Generative AI copilot, Joule. This will help customers better understand upcoming regulatory changes that impact SAP products and business processes. It evaluates a vast number of regulatory updates, putting them into the context of a customer’s business and their SAP solutions. It also provides impact analysis across SAP products and solutions to help customers maintain compliance and run their day-to-day business without disruption.
GL: What are clients telling you when it comes to SAP AI capabilities?
WS: We are seeing our customers solve problems using SAP’s AI capabilities that they haven’t been able to solve before. Let me share an example – before implementing our AI-based solution, Accenture could not achieve more than a 30 percent automated match rate and wasted time manually entering over 250,000 entries annually.
Using SAP’s AI-powered cash application solution, Accenture’s accounts receivable department uses AI to automatically match incoming payments to their corresponding invoices and client accounts. By applying an AI-based matching model in a pilot project, Accenture almost doubled their automatic match rate – going from 30 percent to 54 percent – and significantly reduced peak-time processing for month-end, quarter-end, and year-end closing.
GL: Do you have other pilot success stories?
WS: Take the manufacturing industry. A single manufacturer has hundreds of trucks that arrive daily at production facilities without anything digital in their manufacturing system. When the truck arrives at the production facility, the driver brings the paper delivery notes to manually check in with the facility manager. Since the manufacturer has no information about what is on the truck, someone needs to sit down and key in the information from the paper delivery notes.
Generative AI can transform the process. The machine can read the document automatically and can interpret the information on the document. Then it feeds the information that it captured from the document into an API or into an intelligent document processing engine that is part of the manufacturer’s overall transportation management system.
In a customer pilot at just one facility of a manufacturer, with every document, they estimate that they can save 30 minutes of the time of a person that would have to manually key in that information. That adds up with hundreds of trucks at each facility.
The labor savings alone is close to $1m a year for just the pilot facility. In the U.S. alone, trucking accounts for “72.2 percent of tonnage carried by all modes of domestic freight transportation, including manufactured and retail goods” so the orders of magnitude here are significant.
GL: How has the messaging changed for SAP users – if at all – when going to market with these offerings?
WS: SAP has remained steadfast in its devotion to AI development. Developers stand to reap the benefits as Generative AI makes development more accessible and efficient for individual programmers and enterprise teams, but the rapid pace of technological change has driven up global demand for skilled developers who can incorporate AI into their business processes. SAP recognized this need and unveiled new, role-based certifications and learning resources on our SAP Learning site in order to help meet the ever-changing AI landscape.
In addition, tools like SAP Build Code offer AI-powered productivity tools for developers and streamline cooperation with business experts using SAP Build. This tool also draws on the power of Joule to further boost productivity, embedding code generation capabilities for data model, application logic and test script creation.
And with the new AI Foundation on SAP BTP, developers have a one-stop-shop to create AI- and Generative AI-powered extensions and applications. This includes everything developers need to start creating business-ready AI tools on SAP BTP.
GL: How do SMEs fit into the current strategy, and is the messaging for them different in any way?
WS: Even small companies need enterprise grade technology. For example, Joule makes for a much simpler user experience. This provides an easy, familiar way to interact with SAP applications. It also gives SAP an opportunity to engage a further set of customers.
GL: How does SAP ensure all clients are reassured and also compliant with the ethics of AI use?
WS: SAP’s AI Ethics Steering Committee aligns with, and obtains, insights from an external AI Ethics Advisory Panel of cross-disciplinary experts. Reviews occur at every stage of use case development to ensure data privacy and bias mitigation. We also strive to ensure users (customers) themselves are involved in the content development process, and do not use their data to train or improve LLMs, remaining steadfast that client data does not leave the organization. And we’re also transparent with our customers about our models so they can also use their industry-specific knowledge to reduce bias further.