SAP made bold promises about artificial intelligence at its Sapphire conference last year: a Knowledge Graph, Joule Studio, and an AI Agent Hub would all ship by the end of 2025. Today, those tools are technically available, but the rollout has been slower than anticipated. Adoption of Joule Studio in particular has lagged, and the company is already announcing a major revision — version 2.0 — to address customer feedback.

“Joule Studio adoption has been minimal compared to what we’d like,” said Manoj Swaminathan, SAP’s chief product officer for Business Suite, in a briefing ahead of this year’s Sapphire. The tool “was limited to content-based experiences,” he explained. “Anytime more complex agents were involved, it had limited capabilities.”

The core issue, according to SAP’s chief AI officer Jonathan von Rüden, was that SAP initially favored ease of use over power in its architecture. “People wanted to see more pro-code flexibility,” von Rüden said in an interview at Sapphire 2026. “We had gone with a low-code approach. You could give it extension points and tools, but you couldn’t touch the core of it. Now you can build a custom agent, connect it to your own GitHub.”

Customers arrived with “big plans” but quickly encountered limitations. The original Joule Studio didn’t support hard rules, approval gates, or the kind of agentic workflows that enterprises require. “What people want is agentic flows with clear gates and workflows and subagents,” von Rüden said. “Old Joule didn’t provide that. Now it’s all baked together.”

What actually shipped

Joule and the AI Agent Hub are now generally available, though the hub is already getting what von Rüden calls a “massive revamp” with version 2.0. The Knowledge Graph is live and has expanded beyond its original scope. Initially used for building Joule skills, it now feeds context directly to AI agents so they can “figure out how to call something dynamically,” von Rüden explained.

Joule Studio, however, is still in early customer adoption; general availability is expected in the third quarter — roughly a year behind the original target. Joule Work, the new engagement layer announced this week, isn’t expected until the second half of this year.

The slow pace reflects a broader challenge many enterprise software vendors face: bridging the gap between hype and practical, secure, scalable deployment. SAP’s first-generation tools were deliberately cautious, focusing on low-code to make AI accessible to business users. But that trade-off meant enterprise developers found the platform restrictive. “They wanted to build custom agents with their own code, their own frameworks, and integrate with their existing development workflows,” said an analyst who tracks SAP’s AI strategy. “SAP is now responding to that demand.”

What’s different in 2.0

The revamped Joule Studio addresses the gaps that held back adoption. Beyond the new pro-code flexibility, developers can now build with popular agent frameworks like LangGraph and AutoGen. Crucially, agents will have a native understanding of SAP’s proprietary code and data models — something generic tools cannot replicate. This means an agent built on top of SAP’s Knowledge Graph can reason about business objects, workflows, and transactions in ways that external large language models cannot.

It’s an evolution, not a reset, according to von Rüden. “The first runs were geared toward automation,” he said. “Now agents need to bring optimization and intelligence as well.” Customers, including Ericsson, Mercado Libre, and Siemens, are already using Joule agents in production, handling tasks such as procurement approvals, customer service triage, and supply chain exception handling.

To accelerate adoption, SAP is rethinking how it gets AI tools into customers’ hands. Joule Desktop, launching this week, lets individual users build automations without going through IT — a bet that grassroots adoption will move faster than centralized rollouts. This mirrors trends seen with low-code platforms like Microsoft Power Platform and UiPath’s automation cloud, where business users drive initial adoption before IT steps in to govern.

The broader context is that SAP is under pressure to deliver on its AI promises as competitors like Microsoft, Salesforce, and ServiceNow race to embed generative AI into their platforms. Microsoft’s Copilot for Dynamics 365, for example, directly targets SAP’s ERP customer base. Salesforce’s Einstein GPT and ServiceNow’s Now Assist are similarly aimed at automating enterprise workflows. SAP’s approach — a mix of low-code and pro-code, coupled with proprietary knowledge graphs — is meant to differentiate by offering deeper integration with core business data.

Yet the company acknowledges that the first version of Joule Studio was too limiting. The Knowledge Graph, while powerful, was underutilized because developers couldn’t easily attach custom logic. The AI Agent Hub provided a catalog of pre-built agents, but customers wanted to modify and combine them in ways the original platform didn’t allow. Version 2.0 addresses these gaps by opening the architecture: developers can now define agent behaviors using Python code, integrate with external data sources through APIs, and deploy agents across different environments — from cloud to on-premises.

Security and governance remain top concerns. SAP has built approval gates directly into the agent runtime, so any action that modifies business data must pass through a configurable approval workflow. This addresses the enterprise fear of “rogue agents” making unauthorized changes. “We saw customers freeze projects because they didn’t have the controls they needed,” von Rüden said. “Now the gates are part of the agent definition, not an afterthought.”

The timing of the 2.0 announcement — just as the original tools are reaching broader availability — is unusual but reflects the reality of rapid AI innovation. SAP is essentially skipping a version cycle to meet customer demands. The company is also investing in education: a new “AI Academy” for partners and customers will offer hands-on labs for building agents with Joule Studio 2.0.

Meanwhile, the Knowledge Graph continues to expand. Initially built on SAP’s own data models, it now ingests third-party data sources and can be extended with custom ontologies. This allows agents to reason across business silos — for example, correlating supplier performance data from an SAP Ariba instance with inventory data from S/4HANA and external weather data to predict supply chain disruptions.

Industry observers say SAP’s strategy is sound but execution has been uneven. “They made a lot of noise at Sapphire 2025, but the actual products were half-baked,” said a principal analyst at a technology research firm. “The good news is they’re listening to customers and iterating fast. The 2.0 release shows they understand what enterprises actually need: flexibility, control, and deep integration.”

Still, the delay means some customers have already built parallel AI solutions using other platforms. “We couldn’t wait for Joule Studio to mature, so we used LangChain with our own data,” said an IT director at a large European manufacturer who asked not to be named. “We’ll evaluate the new version, but we’ve already invested months in a custom solution.” SAP faces the challenge of winning back those customers while attracting new ones who are just starting their agentic AI journeys.

SAP’s AI roadmap now extends well beyond the original trio of tools. At Sapphire 2026, the company demoed Joule Work, a new engagement layer that combines chat, document generation, and task automation into a unified interface. Joule Desktop, aimed at individual power users, is already available. And a set of “Industry Agents” — pre-built AI assistants for manufacturing, retail, healthcare, and utilities — is in development.

The company is also embedding AI deeper into its core products. S/4HANA now includes an AI-powered “exception handler” that automatically flags unusual transactions and suggests corrective actions. SuccessFactors’ talent management suite uses AI to recommend learning paths and identify flight risks. And the BTP (Business Technology Platform) now offers APIs for developers to integrate SAP’s AI capabilities into their own applications.

For now, the message from SAP’s leadership is one of humility and acceleration. “We didn’t get everything right the first time,” von Rüden admitted. “But we learned a lot from our customers, and the next generation of our AI tools will be more powerful, more flexible, and easier to use. That’s the promise we’re making this year.”

Source: InfoWorld News