SAP has a big vision for AI agents. It wants them to run business processes, connect data across enterprise systems, and eventually be priced on the outcomes they deliver.
While this might sound great on paper, CX leaders would be forgiven for wondering where the customer service wins are.
There's plenty of sizzle, but is there enough steak?
During SAP’s Q2 2026 earnings call, the vendor talked at length about its Autonomous Enterprise strategy, new AI capabilities, and the potential of Joule Work.
Yet when it came to live customer-facing deployments, contact center transformations, or service outcomes, the evidence was thin.
During the call, Christian Klein, CEO and Member of SAP’s Executive Board, made the company's ambition clear:
“In the age of agentic AI, SAP is leading the way.”
That is a bold claim. But much of the AI story shared during the call remains tied to Q3 launches, beta programs, and longer-term promises.
Joule Work Is Coming, Not Here Yet
The centerpiece of SAP’s next phase is Joule Work, a new AI workspace intended to provide employees with a single interface for collaborating with AI agents across SAP’s portfolio.
Klein explained that during Q3, SAP "will also launch our new end-to-end user experience, Joule Work.
"It’s a single entry point and interface across all our portfolio solutions for all tasks where users can collaborate with our AI agents.”
SAP suggested that a salesperson could use the platform to create a “complete data rich customer pitch in just a few minutes.” This has clear CX connotations, especially for sales teams that need a fuller view of the customer.
However, it is not a contact center deployment. Nor is it evidence that SAP is already improving customer service metrics, such as first-contact resolution, customer satisfaction, or average handling time.
The same applies to SAP’s AI Agent Hub, which Klein described as “our command center to discover, manage and govern SAP and non-SAP agents, MCP servers and more.”
Again, the concept sounds impressive. Enterprises will need a governance layer if they are to deploy multiple AI agents across sales, service, finance, and other functions without creating a security, compliance, or data-management headache.
Still, SAP did not provide evidence of broad customer rollout during Q2. Instead, it pointed to future availability and early interest, saying that “the beta programs for our new platform, Suite and Joule Work were immediately oversubscribed.”
Oversubscribed beta programs are encouraging, but they are not the same as proven production outcomes.
SAP Has AI Results, But They Are Mostly Operational
To be fair, SAP was not all talk.
The vendor shared examples of AI delivering tangible results for customers. But these examples sat largely in finance, procurement, data, and ERP modernization rather than customer service.
SAP and Amadeus developed an AI agent that reconciles unstructured payment data, which Klein said was “already clearing around 40,000 incorrect transactions.”