At its annual SAP TechEd conference, SAP announced new capabilities for Joule, its generative-AI copilot, including collaborative AI agents to tackle more complex business tasks, a knowledge-graph solution to ground its AI technology in SAP-specific business semantics, and new tools for developers focused on accelerating business-AI innovation.
Building upon last year’s announcement of SAP Build Code—a pro-code addition to SAP Build’s low-code environment that offers developers the ability to leverage Joule while creating new applications or extensions for SAP solutions—SAP revealed enhancements to SAP Build that streamline SAP S/4HANA extension development, expand access to ABAP Cloud development tools, and add new generative-AI capabilities.
Additionally, the software company announced embedded data-lake features for SAP Datasphere, allowing organizations to better manage and analyze data across cloud and hybrid environments, and unveiled a knowledge graph engine for SAP HANA Cloud. The addition to the database-as-a-service helps users uncover complex and meaningful relationships between data points that would not be otherwise discoverable with traditional data modeling tools.
Finally, SAP announced that Joule will be available to all customers on SAP S/4HANA Cloud, public edition, by year’s end; the digital copilot will also be available in SAP HANA Cloud, SAP LeanIX, SAP Sales Cloud, and SAP Signavio, with availability in SAP Concur and SAP Service Cloud planned for early 2025.
Collaborative AI Agents for Joule
In introducing collaborative multi-agent systems to Joule, SAP aims to enhance the solution and expand its capabilities to support 80% of SAP’s most-used business tasks by year’s end. These specialized AI agents will tackle specific tasks and communicate with one another on complex workflows, working both autonomously and collaboratively to break silos across organizations.
“We are infusing Joule with multiple autonomous AI agents that will combine their expertise across the business functions to collaboratively accomplish complex workflows,” said Muhammad Alam, Head of Product Engineering at SAP, in a media briefing. “This will free workers to collaborate in areas where human ingenuity is best suited.”
SAP detailed two finance-specific use cases for these collaborative AI agents, which will be prompted to complete tasks by human employees communicating with Joule:
- Dispute management: Autonomous AI agents can be deployed to help employees analyze and resolve dispute resolution scenarios, including incorrect and missing invoices, unapplied credits, and denied or duplicate payments.
- Financial accounting: Autonomous AI agents can be deployed to help employees to streamline key financial processes by automating bill payments, invoice processing, and ledger updates while quickly addressing inconsistencies or errors.
Developers will also be able to create their own autonomous AI agents via Joule to tackle specific business challenges, beginning early next year, according to SAP.
SAP Build Additions to Enhance Application Development
Bringing generative-AI capabilities to the SAP Build platform, SAP aims to empower developers to further extend SAP solutions and leverage the digital copilot’s unique knowledge of SAP data and processes to do so. In enhancing SAP Build, SAP is framing the platform as the extensibility hub for all SAP applications, also centralizing lifecycle management and governance for all SAP projects.
Code explanation, code completion, business object generation, unit test generation, documentation search, and instant assistance from SAP Help and SAP Community are among the new generative-AI capabilities available to developers, which will reduce development time for Java and JavaScript developers. The digital copilot can also provide workflow automation and approval recommendations. SAP is making it easier for developers to extend Joule, creating and managing custom AI skills to their work within Joule Studio in SAP Build, available by year’s end.
Through a new extensibility wizard, now in general availability, developers can directly access SAP Build from SAP S/4HANA, public edition, deepening connectivity between the two.
ABAP developers and fusion teams will have seamless access to ABAP Cloud development tools from SAP Build, through an SAP Build lobby that will serve as the central entry point for all development tools and feature access to ABAP. Streamlined navigation between ABAP, Java, JavaScript, and low-code tools will preserve full business context more effectively and subsequently empower full-stack application development.
Empowering Data and Analytics
With its announcement of a new embedded data lake for SAP Datasphere, available by year’s end, SAP is expanding its previously available data lake service (via SAP HANA Cloud, data lake), enabling enterprises to store large amounts of data while retaining valuable business context.
Enabling users to import and integrate data at scale, the updates will also ensure seamless integration from both SAP and non-SAP sources, preserving the original structure and context of business data (including structured, semi-structured, and unstructured data).
Key features of this data lake option, which expands the business data fabric architecture of SAP Datasphere while complementing existing storage solutions, include:
- Integrated object store for scalable storage
- Spark compute for transforming and processing data more quickly in the object store
- SQL on files functionality for accessing and querying data without duplication
Generally available in Q1 2025, the knowledge graph engine for SAP HANA Cloud will improve users’ abilities to map relationships and context across their organization’s entire data landscape and simplify the data modeling process.
Based on the industry standard Resource Description Framework, the knowledge graph engine stores information in three parts—the subject of the data, the object it’s related to, and the nature of their relationship—and “organizes data into a web of interconnected facts, making it easier to see how different pieces of information relate to one another,” according to SAP.
This approach can provide context to large language models (LLMs) and reduce AI-generated hallucinations in data; knowledge graphs are more accurate than vector engines when dealing with structured data in the context of LLMs, improving the accuracy and relevance of generative-AI output.
Finally, with a real-time risk analysis feature in SAP Analytics Cloud, referred to as the compass feature and now in general availability, business users can model complex risk scenarios and calculate the probability of potential outcomes. Building upon previous Monte Carlo simulations in existing SAP tools, business users can now more efficiently predict or forecast business scenarios using enterprise data, without advanced statistical skills.