Joule, the SAP Agent Thriller

Drawing the line between operational ERP and GenAI Agents

The following quote is credited to the Austrian philosopher Ludwig Wittgenstein (1889-1951): "The limits of my language are the limits of my world." While this may sound harmless or even interesting from a philosophical point of view, it could be revolutionary—or even dangerous—in the field of GenAI, LLMs, and AI agents. Why?

Using the current SAP vocabulary, ERP is built and operated sequentially. S/4 is better than ECC 6.0, which is better than R/3. The SAP user orchestrates. The ERP system works. There is an extensive division of labor between humans and machines, which is defined by SAP's work and specifications.

However, in an agile, open, composite ERP managed by AI agents, the division of work and tasks is unclear and probably irreversible. Rather than automating, AI agents should orchestrate independently. With this instruction, AI agents will quickly surpass the capabilities of our ERP language.

In a composable ERP, AI agents can develop independently and dynamically. To paraphrase Ludwig Wittgenstein: AI agents can learn new ERP languages (consolidation, harmonization, and orchestration) independently and build a composable ERP that extends far beyond our own borders. Perhaps SAP will become the new DeepL of the ERP scene with Joule and GenAI Hub from BTP.

Composable ERP as an SAP playground

SAP has often and intensely spoken about the intelligent enterprise in the past. SAP provides many IT tools for implementing an intelligent business structure and organizational processes, such as the Hana Graph and Vector engines, the BTP GenAI Hub, and Joule. However, SAP has yet to define positions and assign tasks. What should AI agents be allowed to do? Where should LLMs, or large language models, look? Are there limits for the GenAI Hub in Composable ERP?

Many years ago, AI researchers experimented with neural networks using the well-known board game Go. Initially, the neural network—which we would probably call an LLM nowadays—was familiarized with the rules of Go and trained using existing Go literature. Soon after, the Go computer began playing much better than the world's best Go player. The machine-made moves that Go experts considered absurd and counterproductive, yet the machine always won in the end.s

In a second round, the AI researchers fed the neural network only the fundamental rules of Go and let the machine play against itself continuously. The Go computer learned and perfected its Go language. In doing so, it surpassed the entire body of Go literature from the past thousand years. Through back-reference, the Go computer surpassed human capabilities.

Has SAP considered this option for AI agents moving, developing, and optimizing in a composable ERP? In theory, SAP Joule could use the building blocks of a composite ERP system to develop entirely new business management systems and organizational structures.

IT, and ERP in particular, are in the midst of an agent thriller. AI agents, such as SAP Joule, have arrived to orchestrate ERP work, optimizing business structures and processes. These agents create the composable ERP. Further development, research, and version changes are carried out by AI agents. Hopefully, humans will remain beneficiaries or merely spectators.