SAP SE today announced a shift from promising agentic artificial intelligence autonomy to defining a path for users to interact with agents for work outside apps, not just within them.

During SAP Connect 2026, the company said it is making Autonomous Enterprise architecture, introduced at Sapphire in May, generally available this month alongside Joule Work and Desktop, with an expanded suite of autonomous assistants and agents.

“When we talk about autonomous enterprise, we realize that it’s going to be a journey for the customer, and it’s not something you draw a line to say starting tomorrow will be autonomous just because you now have these agents and assistants,” Manoj Swaminathan, SAP’s president and chief product officer of Autonomous Suite, told SiliconANGLE in an interview.

Joule, SAP’s generative AI assistant, debuted in 2023 and has become the company’s autonomous workhorse. Today, the company announced numerous upgrades, making it a powerful automation, productivity and development tool that acts as a conversational interface layer for customers that works alongside apps rather than within them.

According to Swaminathan, “autonomous” doesn’t mean simply handing everything to agents and then letting them go; it means providing humans the tools to observe, describe, and trust agents before delegating processes to them. This starts with the opportunity to control, govern, and understand how agents will work with their data, apps, and enterprise context before setting agents loose, and with the transparency to see and audit their activity.

“First principle is it’s AI on apps as opposed to AI in apps,” Swaminathan explained.

Front and center, Joule Work provides the framework to bring AI assistants to customer workflows, allowing autonomy to work with both SAP and non-SAP applications. A user looking at a screen has already experienced AI agents working within apps, where they can open up a word processor and have an agent write them a memo, or an email program, and have an agent look up past conversations. But having an agent understand enterprise resource planning and connect finance documents and email conversations without needing to jump between the two becomes SAP’s domain.

For example, Swaminathan gave an example of total corporate spend. Rather than treating the question as belonging to one application, Joule could determine that answering it requires information from ERP, travel and expense systems and the procurement platforms. It could then combine those records, present the result as one business answer.

This interface system is called Joule Desktop and it lives where the customer does with their apps and grabs the same enterprise information. It provides an intelligent layer that can perform duties that deal with the work itself rather than bind them to in-application resources.

“Joule is no longer a natural language chat client for us; it is a full-screen experience,” Swaminathan said.

With the framework’s general availability, Jule will receive updated agents for finance, supply chain, spend, workforce/customer experience, and other industry specializations across the autonomous suite. This will give customers access to deep capabilities and agents with domain expertise, knowledge and business acumen that can touch enterprise data.

Moving from AI models to autonomous enterprise

According to Swaminathan, enterprise customers bring their data and cloud lakehouses; SAP organizes that data and provides it with a structured knowledge graph. General-purpose models and agents from OpenAI Group PBC and Anthropic PBC cannot provide this when simply pasted on top of enterprise data. ERP data, processes, semantics, policies and relationships all matter – this is the “secret sauce” that SAP brings to the table.

SAP isn’t alone in this trend. Salesforce Inc. and ServiceNow Inc. are also driving an industry trend toward headless applications, context layers and agentic control planes. Enterprise AI is becoming less about adding copilots to applications themselves and more about building an intelligent layer on top of the customer experience.

Swaminathan stressed that the opportunity for SAP customers is the ability to declare intent; the agent decides which systems to involve and then gets the work done. The human always stays in the loop, retains governance, auditability, and transparency and agents are treated like employees in the system, including security, systems of record, and execution. Agents are another layer on top of the old application interface that act as intelligent conversational tools on behalf of humans.

“Autonomous enterprise is not a vision anymore,” Swaminathan said. “It is real, with general availability in October.”

SAP will continue using AI units as the primary commercial monetization for these capabilities. The model will be primarily consumption-based rather than purely subscription-based, with agent inference consuming units and SAP providing customer visibility into that usage.

The Joule updates and other autonomous enterprise capabilities will become available throughout October and customers can already see agents in SAP’s Discovery Center.

Image: SiliconANGLE/Microsoft Designer