SAP is extending its SAP HANA foundation for real-time, in-memory data processing with Joule, its unified AI assistant, and a broader business AI platform that combines enterprise data, process context, governance and security.
SAP’s emerging autonomous-enterprise vision goes beyond placing generative AI atop existing workflows. It envisions connected agents working across five domains – finance, supply chain, HR, spend management and customer operations – while keeping humans in the loop. In effect, SAP HANA made data immediately actionable, while Joule serves as a unified engagement layer for SAP, its AI agents and underlying applications.
These AI agents aim to make enterprise processes increasingly proactive, coordinated and autonomous.
During his welcome keynote at the SAP NOW AI Tour in Singapore, Liher Urbizu, SAP president and managing director for Southeast Asia, offered an example of what business AI can deliver for financial reporting functions.
“Your financial reports cannot be 80-percent accurate sometimes. They need to be 100-percent accurate all the time.”
In essence, SAP’s business AI is about generating trusted answers grounded in an organisation’s enterprise data, drawing on SAP’s 50 years of process and data-domain expertise.
Liher explained, “What we do at SAP is we train AI to understand SAP business processes, to understand critical business data with its semantic richness, and we build the risk and compliance, identity, access and security around it, so that it becomes accurate, fit for business purpose and, most importantly, safe and secure to use.”
SAP’s knowledge graph
Knowledge-graph technology turns data into a network of entities and relationships—for example, a customer linked to its contracts, orders, suppliers and more. The graph adds meaning and context that conventional spreadsheets cannot capture as effectively.
This makes the knowledge graph significant. As Liher put it, “Our value is not the LLMs [large language models] … we have opened up to all publicly available LLMs for a simple reason.
“Our value is in building an integrated common semantic layer, called the knowledge graph, that connects all the dots between data and process so AI is fit for business …”
It also allows AI agents to operate more safely and accurately across an organisation’s technology landscape.
Renewed industry attention on knowledge graph technology has emerged as businesses explore and adopt agentic AI. LLMs can generate fluent responses, but the next phase of AI requires agents to understand an organisation’s processes, data definitions and business rules.
Autonomous enterprise
According to Liher, as many as 34,000 customers are already using AI. “The opportunity is immense. Southeast Asia is one of the hotbeds of AI adoption. This is what we see when we look at internal consumption data from our customers in SEA compared to the region.”
SAP’s data also revealed that there is no industry or business process where AI lacks potential applications.
ComfortDelGro’s CIO Dirk Bell joined Liher on stage to discuss the group’s vision of building a unified digital core across 13 countries and 26,000 employees using S/4HANA and SuccessFactors. These deployments aim to harmonise processes and standardise data so management can make decisions based on consistent information across its three main pillars: public transport, private point-to-point transport, and testing and certification.
Our value is in building an integrated common semantic layer, called the knowledge graph, that connects all the dots between data and process so AI is fit for business.Liher Urbizu
Dirk also shared, “I think one of the key developments that we’ve done over the last couple of months is transforming our call centre functionalities by replacing a lot of the low-value calls to agents, so that operators can actually really spend more time and quality time in the value-added conversations with the customers.” Over the past seven months, the company has been piloting the initiative and is now operationalising it in the UK, Australia and Singapore.
Singapore Airlines’ (SIA) divisional vice-president for enterprise strategy and architecture, Satyajit Deshpande, later took to the stage to frame digital as the strategic foundation that frees frontline staff from mundane tasks. AI initiatives that began a few years ago led to an AI blueprint last July, through which SIA identified more than 550 use cases; about 160 are already live. A rearchitected chatbot has doubled customer-satisfaction scores, while an Agent Builder tool now enables SIA’s business users to build an estimated 3,000 agents for their own workflows.
Satyajit acknowledged SIA’s extensive SAP footprint. This led to the creation of Project Polaris, which he described as SIA’s reimagining of its entire ERP landscape to future-proof the organisation.
SIA wants to deploy AI across all functions where it makes sense and generates business value, he said. For example, it is working closely with SAP in sourcing, contract management and supplier performance to build capabilities that uplift the procurement function.
All in all, both SIA and ComfortDelGro stress strong foundations like unified data, well-defined processes, and governance, to augment their people’s skills and capabilities. This aligns with SAP’s vision of an autonomous enterprise, where AI agents take on the heavy lifting so human potential has more room to shine.
(This journalist was invited by SAP to attend this event in Singapore)