SAP’s language model assistant Joule has been part of the SAP narrative since 2023, but the framing shifted substantially in the twelve months leading into mid-2026. The headline word is now “agentic.” Understanding what that means in practice, and what it does not mean yet, is useful before a utility team commits time and budget to an AI roadmap built on the SAP stack.
What “Agentic” Actually Means
A traditional AI assistant responds to a prompt. You ask a question; it returns text. An agentic system goes further: it can plan a multi-step task, call tools or APIs, read from one part of a system, write to another, and complete the task with minimal human involvement at each individual step.
For a utility, the practical difference is scope. A non-agentic Joule might surface a customer’s payment history when a call-center agent asks. An agentic Joule can receive a customer query, look up the account, identify a billing exception, initiate a correction workflow, and send a resolution message, all without the agent manually touching each step. The IS-U or S/4HANA Utilities system remains the system of record; Joule orchestrates the sequence around it. For more on how AI sits alongside the ERP rather than replacing it, see our guide to integrating AI with SAP and Oracle for utility operations.
Joule’s Scale Across the SAP Portfolio
The agent and skill counts SAP publishes have moved fast and the numbers carry a date stamp because they change. At SAP TechEd in November 2025, SAP set a Q4 2025 target of roughly 40 agents and 2,100 Joule Skills. By the Q1 2026 release summary published in April 2026, SAP reported 30 or more specialized agents and more than 2,500 Joule Skills. At SAP Sapphire in May 2026, SAP reframed the count under a new Autonomous Suite umbrella, citing more than 200 agents and more than 50 assistants.
Not all of those apply to utilities. Utilities-specific coverage is a subset of the portfolio, and teams should evaluate which agents map to their actual processes in IS-U or S/4HANA Utilities before drawing conclusions from headline totals.
The Utilities Customer Self-Service Agent
One concrete utilities deliverable is the Utilities Customer Self-Service Agent. SAP reported it reached general availability in Q4 2025. It is designed to handle inbound customer inquiries, common account tasks, and routine service requests without routing every interaction to a human agent.
For a utility running SAP IS-U or S/4HANA Utilities as its customer information system, this agent pulls account data from the CIS and responds through a conversational interface. The CIS remains authoritative; the agent operates in front of it. The practical question for any utility evaluating this is whether the customer journeys the agent covers align with the service types that generate the most call volume in their operation.
Joule Embedded in Asset Management
Separately from the customer-facing agent, SAP embedded Joule in the Asset Management module in the S/4HANA 2025 on-premises release in October 2025. The embedded capabilities include natural-language data retrieval and work-order creation, meaning a maintenance technician or planner can describe a task in plain language and have Joule structure the work order in the system.
For asset-intensive utilities managing transmission and distribution infrastructure, this addresses a real friction point: the time and training required to navigate SAP Plant Maintenance screens to create and update work orders. Natural-language input lowers that barrier, though the underlying data model and configuration quality in PM still determine whether the results are accurate. The SAP S/4HANA Utilities and BTP platform context is useful background for teams thinking about where these embedded capabilities sit in the architecture.
The RWE Offshore Wind Example
At SAP Sapphire in May 2026, SAP cited RWE’s offshore wind operations as an example of agentic AI reducing unplanned downtime. SAP framed this as exploring or introducing agentic AI in the context of their broader Industry AI coverage across 26 industries, not a fully deployed production system. RWE is an early reference customer in a program SAP is actively developing, and utilities evaluating similar use cases should treat it as a directional signal rather than a proven deployment template.
The underlying concept, using agentic AI to detect anomalies and route maintenance actions in grid or generation assets, is a logical fit for utilities that have already standardized asset data in SAP PM and have reliable sensor feeds. The quality of the underlying data in the ERP is the main gating factor.
E.ON as a Real Production Adopter
A more quantified reference point is E.ON. Reported in June 2026, E.ON has been deploying SAP S/4HANA-based AI for predictive maintenance and customer service and reported a 77% reduction in IT downtime over five years. E.ON is one of the larger European energy distributors, and the result points to what a mature, sustained SAP AI program can produce over a multi-year horizon rather than a short pilot.
The distinction from the RWE example is that E.ON’s reported outcomes are from an operational deployment at scale, not a capability preview. Utilities in mid-migration from IS-U on ECC to S/4HANA Utilities can use E.ON as a concrete benchmark for what the destination state can deliver. For a comparison of where SAP and Oracle currently stand on this kind of capability, see our Oracle vs SAP utilities comparison.
What Joule Does Not Do Yet
Several capabilities remain forward-looking. The SAP AI Agent Hub and a new managed Joule Studio are both targeted for general availability in Q3 2026, which is when utilities teams would be able to build and govern their own custom agents in a managed environment. The original Joule Studio agent builder reached GA on December 19, 2025, giving developers an early path, but the managed enterprise-grade tooling is still ahead.
The Anthropic partnership announced May 12, 2026 will embed Claude in the SAP Business AI Platform via the Model Context Protocol. Utilities are named among the target industries in that announcement, but no utilities-specific agent under the Anthropic partnership has shipped yet. It expands the model choice available within the SAP AI Platform rather than delivering a ready-made utilities workflow today.
For utilities teams doing roadmap planning, the pattern is: the customer-facing and asset-management capabilities are available now, the custom-agent tooling arrives in Q3 2026, and the multi-model capabilities that follow from the Anthropic partnership are later still. Building the AI roadmap in that sequence, starting with the GA agents and working toward custom builds, is a more grounded approach than planning around capabilities that are still targets. The SAP IS-U overview covers the platform context for teams earlier in their evaluation.