Reliable automation is possible primarily for verifiable administrative operations and supporting tasks, where AI searches, extracts, cross-checks or prepares a draft, but does not determine the outcome for a person. Candidate selection, employee evaluation, promotion, dismissal and the allocation of opportunities require separate risk classification, valid criteria, group-level control, logging, real human review and the right to appeal. Some scenarios, including emotion recognition in the workplace from biometric data in the EU, may be directly prohibited. The article shows how to choose automation scenarios and which changes to process, data, control and accountability are needed before deployment.
The boundary runs not between a “simple” and a “complex” requirement, but between parameterizing behavior that SAP already provides and creating new behavior. FI/CO owns financial semantics, the standard process, and Customizing; development begins where program logic, a published extension point, a new interface, or an external application is needed. The article provides a decision tree, a role distribution, and a conditional walkthrough of a journal entry posting requirement.
An execution plan does not explain a delay by itself: it shows the way the optimizer chose to process the data. To find the cause, you must match the estimates against the metrics of the specific execution, find the first significant cardinality gap before the growth in work, account for repetitions of operators, and check I/O, temporary data, and waits outside the plan. The article gives a step-by-step analysis method for PostgreSQL, SQL Server, Oracle Database and MySQL 8.x.