Freedom in Walldorf

*Freedom in Krähwinkel* is a musical farce in two acts by my local poet Johann Nestroy (1801–1862). The play was written in 1848, the year of the 1848–49 Revolution, and premiered on July 1, 1848, as a benefit performance for Nestroy at the Carl Theater in Vienna, Austria. After the revolution breaks out, the fictional town of Krähwinkel also demands its own „freedom.“ The citizens form vigilante groups and committees, but they don’t know what to do with their freedom. In the end, the old order prevails out of fear of chaos.

AI is an IT revolution, and at present, very few companies and users know what to make of this revolution. Who will stay, who will have to leave, and who will benefit? The theory that SAP’s AI-driven platform architecture—centered on AI, Agentic AI, the Autonomous Enterprise, and the Knowledge Graph—will grant SAP’s existing customers independence is proving to be a narrative with two sides to it.

In SAP’s keynote presentations, the Knowledge Graph is portrayed as the semantic „Babelfish“ that automates, maps, and makes machine-readable the outdated ERP data model, which consists of over seven million data fields and tens of thousands of entities. By translating relational tables into ontologies and a network structure of nodes and edges, the goal is to prevent language models in the business-critical general ledger from producing dangerous “hallucinations.”.

However, anyone who believes that Walldorf is handing its existing customers the key to breaking free from vendor lock-in is deluding themselves.

The Babel Fish is a fictional creature from the novel *The Hitchhiker's Guide to the Galaxy* by Douglas Adams. Douglas Adams derived the name „Babel Fish“ from the biblical story of the Tower of Babel and the confusion of languages in Babylon, as described in the Book of Genesis.

The Knowledge Graph and the Business AI Platform (BAIP) are primarily intended to serve as a new framework for SAP’s ERP ecosystem. Access to the SAP-managed Knowledge Graph remains tied to paid use of the Business Technology Platform (SAP BTP) and the Business Data Cloud (SAP BDC). Anyone wishing to activate the semantic context of their company data will be directed to the Walldorf “toll booth” via usage-based Capacity Units, token fees, and the restrictive SAP API Policy enacted in April 2026.

Instead of liberating customers, SAP is attempting to lock away the business context—the true crown jewel of every company—behind proprietary interfaces. The clause enshrined in Section 2.2.2 of the API Policy explicitly regulates direct, unlicensed access to SAP data by third-party AI agents, which user associations such as the DSAG view as an attempt that raises antitrust concerns.

For existing customers, liberation from the ERP giant’s stranglehold does not come from resignedly accepting leased cloud models, but rather from the consistent adoption of open, vendor-neutral standards from the field of computer science. Mathematical graph theory and the specifications of the W3C Consortium—the Resource Description Framework (RDF), the SPARQL query language, and the Web Ontology Language (OWL)—are inherently open-source and vendor-agnostic.

When an existing customer chooses not to store its business-specific taxonomies and ontologies in the BTP, but instead builds a universal „Enterprise Semantic Backbone“ based on open and sovereign open-source infrastructures such as SUSE, Red Hat, or Boomi, they undermine SAP’s monolithic claim to power from within. Such an open knowledge graph completely decouples cognitive logic and business understanding from the underlying application software.

Composable ERP and AI-powered „Vibe Coding“ are creating a disruptive dynamic. When business rules, data relationships, and process paths are formally defined in an RDF-based ontology, modern language models and specialized AI platforms—such as Nova Intelligence’s Cleanfield concept—can generate the required application code, the Fiori interfaces, and the validation logic directly from the semantic model.

In this scenario, existing customers no longer need to maintain Z modifications in the ABAP namespace. They can have their extensions programmed by autonomous agents on clean, decoupled platforms, while the S/4 core is downgraded to a pure, interchangeable transaction server for the general ledger. The DSAG Investment Report 2026 supports this trend: While a mere three percent of user companies rely on SAP’s native AI tools, 77 percent of AI-active companies are already using independent language models and third-party platforms to defend their digital sovereignty.