Getty Images/iStockphoto

SAP focuses on developer efficiency in Joule, Build update

At SAP TechEd, the vendor unveiled new functionality for its generative AI copilot, Joule, as well as ABAP capabilities in its Build low-code/no-code platform.

SAP developers are getting more AI-centered functionality and capabilities that are aimed at helping them increase productivity and work more collaboratively.

At SAP TechEd this week, SAP unveiled developer updates for Joule, its generative AI copilot, and SAP Build, its low-code/no-code platform. For Joule, the vendor added autonomous agents to help developers create workflows that span enterprise systems, as well as access to Knowledge Graph to help generate more accurate and business-relevant answers. To Build, the vendor added an Extensibility Wizard so that developers can more easily access the platform from S/4HANA Cloud Public Edition. Developers can also now work with Advanced Business Application Programming (ABAP), SAP's programming language, in Build.

The new capabilities are aimed at striking a balance between standardization and customization so that SAP software can meet customer needs, said Muhammad Alam, SAP head of product engineering, at a media briefing prior to SAP TechEd.

"We want to deliver a seamless suite-like experience across our entire solution portfolio," Alam said. "We want to simplify our customers' lives, driving efficiency and enhancing business processes with the right AI technology and meaningful use cases."

One analyst said the tools could be useful for SAP developers, but they should be wary of overhyped promises around capabilities like AI agents -- and it will be up to the developers to decide how much they will help.

We'll have to see how customers view these agents in practice. They will be useful, but there's a reason why certain processes haven't been automated up until now.
Jon ReedAnalyst and co-founder, Diginomica

"We'll have to see how customers view these agents in practice," said Jon Reed, an analyst and co-founder of Diginomica, an enterprise industry analysis firm. "They will be useful, but there's a reason why certain processes haven't been automated up until now."

Moving developers to where AI is going

The agents can make SAP developer tools more flexible, so responses can change as business conditions change, according to J.G. Chirapurath, SAP chief marketing officer, in a briefing with TechTarget Editorial prior to SAP TechEd.

"You want [the agents] to orchestrate behind the scenes, build these bots with these agents to collaborate with each other, come back with the answer and the outcome that you're looking for," Chirapurath said.

For example, a financial accounting application could use autonomous AI agents to streamline key financial processes by automating tasks such as bill payments, invoice processing and ledger updates, as well as identifying inconsistencies or errors, he said. Customers could also use autonomous AI agents to analyze and resolve dispute resolution scenarios, including incorrect and missing invoices, unapplied credits, and denied or duplicate payments.

SAP Knowledge Graph, a data management capability that was included in SAP Datasphere in March, anchors Joule copilots to a company's business data to deliver reliable and context-aware business insights with fewer inaccuracies, according to Chirapurath. Knowledge Graph maps and correlates the relationships and context of data across enterprise systems.

"It understands not just the corpus of data, but can also understand the meaning of what that data is," Chirapurath said. "When you ask it a question, it can gauge relevancy and can gauge the speed in which it can answer."

For example, he said, if a user asks Joule how many erroneous invoices went through a finance system for a specified time frame, it must discover the invoices but also determine which are the invoice numbers and how to define erroneous.

Knowledge Graph models data across all systems, including business semantic information, which connects relationships between data, metadata and business processes, Chirapurath said.

The new capabilities in SAP Build are designed to help developers work more efficiently and productively, Chirapurath said. Build originated two years ago as a low-code/no-code development environment, and last year added Java and JavaScript capabilities for higher-level development.

Now developers can access SAP ABAP code directly in Build, and the Extensibility Wizard enables them to build and deploy extensions from S/4HANA Cloud Public Edition, he said.

"The benefit to the developer is productivity -- they don't have to go to different environments for different tooling," Chirapurath said. "But it's not just a development tool, it's also a collaboration tool for developers and business experts."

Joule trained with developer modalities is also now available in SAP Build, enabling developers to use Joule for code generation and explanation, he said.

"Let's say you have 10,000 lines of ABAP code that was written 20 years ago," Chirapurath said. "You can point Joule at it and say, 'Tell me what this code is, what it does, and can you rewrite it into JavaScript.' Joule will then do that for you."

Beware of AI agent automation hype

Diginomica's Reed cautioned that enterprise vendors tend to imply that AI changes things for enterprise development more than it does in practice. However, he added, the inclusion of ABAP capabilities in Build and the deeper integration of Build into S/4HANA public cloud could be useful.

Customers need to be wary of any overhyping of AI agents, he said, as automating workflows can be problematic without human supervision. One example of this is SAP's use case of using AI agents to automate dispute resolutions.

"I'm not so sure that, as a customer, I'd be thrilled with having AI trying to resolve finance or credit disputes," Reed said. "Of course, it depends on how SAP has designed this, and how easy or hard it is to escalate to humans."

Also, if the large language model underpinning generative AI bots gets one of the agent tasks wrong or doesn't understand what the user wanted, adding more autonomous steps can compound the error, Reed added.

SAP customers could be able to use the Knowledge Graph functionality as an alternative to vector databases in retrieval-augmented generation scenarios, he said. Other vendors are working on similar technology, but SAP appears to be the furthest along with Knowledge Graph.

"In particular, this is promising for the accuracy and relevance of GenAI output in the enterprise because a knowledge graph is better at showing context between data points and is also better with a structured transactional type of enterprise data," Reed said. "In theory, this could be a good piece of the value proposition for [SAP's business-focused AI offerings]."

Jim O'Donnell is a senior news writer for TechTarget Editorial who covers ERP and other enterprise applications.

Dig Deeper on SAP development and programming languages

Search ERP
Search Oracle
Search Data Management
Search AWS
Search Business Analytics
Search Content Management
Search HRSoftware
Close