Academy module

When You Do Not Need an Agent: Rules, Situation Handling, ML or LLM in S/4HANA and SAP BTP

As of 2026-11-05

Edited by Cédric Mary, publication director — 27 years in the SAP ecosystem. How these pages are produced, and where automation is used.

Most requests for an AI agent in an SAP project hide a simpler need. This module gives a four-rung ladder: a rule, a notification, a prediction, a generation. It compares Situation Handling in S/4HANA, Intelligent Situation Automation, decisions in SAP Build Process Automation, ISLM and machine learning, and Joule skills and agents on cost, determinism, audit trail and failure modes, using SAP Learning pages, developer tutorials, the SAP Architecture Center AI Golden Path and the SAP AI Core documentation. You build a decision table, prove it returns identical results, and then decide which items of a realistic request need an LLM. The answer is usually one, and a person still sends the result.

What you will learn

Place any requirement on a four-rung ladder (rule, notification, prediction, generation) and justify the rung
Configure a Situation Handling type from a template and name what Intelligent Situation Automation adds
Build a decision table in SAP Build Process Automation and prove it is deterministic with six test cases
Compare cost, determinism, audit trail and failure modes of rules, ISLM, Joule skills and agents
Write a one-page recommendation that keeps rules as tools and limits the LLM to language

Concept overview

The cheapest agent is the one you do not build. Before a Joule agent, a generated answer or a trained model enters an SAP process, a senior consultant asks one question: what kind of output does this step need? If the output is a fixed answer from fixed inputs, a rule gives it for free and the same way every time. If it is a heads-up for a person, a notification does it. If it is a number learned from history, a machine learning model does it. Only if it is language, or a plan that cannot be written down in advance, do you need an LLM or an agent. SAP's own AI Golden Path says the same: its Technology Decision Tree starts by asking whether "process optimization or traditional automation" would solve the problem "more effectively" before any AI is chosen (Technology Decision Tree, as read on 5 November 2026).

This module is the decision module of the track. It compares five mechanisms that exist in an SAP landscape and that are regularly confused:

Prerequisites

An SAP Build Process Automation trial or productive tenant for the lab
Basic knowledge of SAP S/4HANA business roles and the Fiori launchpad
Neighbour modules M369, M373 and M341 read or at hand for the build side

Outcomes

Explain why a compliance control must not rely on a Situation Handling notification
Reproduce the lab: cases 1 and 2 return SO_APPROVER, cases 3, 5 and 6 return SO_MGMNT, case 4 returns an empty result
State what the sources document for Intelligent Situation Automation and what they leave open
Choose between sap-rpt, classic ML with PAL or APL, and an LLM for a given data shape
Name the audit evidence each rung produces and who must switch it on

Full module with the Consultant plan — €19.90/month. The full module adds: the decision framework · the end-to-end scenario walkthrough · the KPI scorecard · the anti-patterns · the code blocks · the knowledge check · the diagrams.

Open in the app →

Open missions

120 live roles and contracts in SAP AI & Joule right now — see the SAP AI & Joule missions · freelance SAP contracts.