What happened on May 12, 2026
At SAP Sapphire, SAP's flagship event, came an announcement that changes the terms of the enterprise AI conversation. SAP and Anthropic made Claude the primary reasoning and action capability of Joule, SAP's AI assistant, embedded in the newly launched SAP Business AI Platform.
This isn't a generic press-release partnership. SAP CEO Christian Klein framed it this way: "The Autonomous Enterprise requires AI that understands business context and acts within the controls organizations depend on." On Anthropic's side, co-founder and President Daniela Amodei added that with Claude on SAP Business AI Platform, "that work happens inside the systems enterprises already invested in" — no need to migrate elsewhere.
How it works: MCP is the connective tissue
Here the news touches directly on the work we do every day. The connection between Claude and SAP systems runs through the Model Context Protocol (MCP), the open standard developed by Anthropic to connect AI models to external data and tools.
With MCP, Joule agents coordinate actions across SAP S/4HANA, SAP SuccessFactors and SAP Ariba, plus third-party systems. It's not a point-to-point integration built ad hoc: it's a standard protocol that lets the AI work inside existing SAP infrastructure. It's the same principle behind how we help companies integrate Claude into their own systems — except here the system is SAP, the most widely used enterprise ERP in the world.
What the agents actually do
The leap beyond a chatbot is that Joule agents powered by Claude perform actions, not just respond. The examples SAP gives: closing the books, answering employee leave requests, rerouting orders to suppliers when something goes wrong.
The areas covered are finance, HR, procurement and supply chain — the operational processes that weigh most in an enterprise's day-to-day. It's a category shift: from "AI that analyzes documents" to "AI that closes a process".
Want to find out where Joule can act in your business?
30 minutes to discuss your specific case.
Governance isn't bolted on, it's native
The point that matters for anyone working in regulated contexts: agents operate "within the same controls that govern human decisions — the approvals, policies, and compliance frameworks already wired into SAP solutions," to quote the SAP announcement directly.
In other words: the agent doesn't bypass internal controls, it works inside them. It's the same security-by-design principle found in Anthropic's approach to its own models — applied here to the business-process layer.
A nuance worth noting
SAP didn't make Claude the only available model. The company maintains an open, multi-model ecosystem approach — Claude is the primary reasoning engine, not an exclusive one. It's a choice consistent with SAP's open-platform strategy, and worth knowing so as not to oversimplify the news.
Why it matters if your company runs on SAP
If your company runs on SAP — likely if you're in manufacturing, retail, pharma, or an enterprise-scale group — this news concerns you directly, it isn't just an update for specialists.
Claude is already inside the tool your teams use every day. The useful question is no longer "should we adopt AI," but "which SAP processes can we hand to an agent, and under what controls." That's exactly the work we do with PA ABS, a SAP Gold Partner specialized in S/4HANA: we combine our expertise in Claude and MCP with certified SAP delivery to bring Joule agents into production. If you want to know where to start, let's talk.