About · Design principles

2026

ETM4S4 design principles

The principles used to shape agents, collaboration, governance and human accountability within the mesh.

The ETM4S4 model is guided by a small number of principles.

Transformation knowledge before document generation

Requirements, decisions, designs, tests, controls and evidence should form a connected model. Documents are useful views of that model, not the only place where project knowledge lives.

Human accountability remains explicit

An agent may prepare, recommend, assess or execute within a defined boundary. It does not make human ownership disappear. Material decisions need an accountable person or governance body.

Autonomy must be earned

Agents should begin with observation, drafting and recommendation. Broader authority should follow evidence of performance, stronger controls and proven recovery mechanisms.

Standard first, exception by evidence

For S/4HANA transformation, fit-to-standard and clean-core principles should be embedded in normal agent behavior. Deviations require a clear rationale, impact assessment and approval.

Assurance must be independent

The agent producing a design should not be its only reviewer. The mesh therefore includes assurance agents for evidence, consistency, segregation of duties, quality gates and risk.

Skills are explicit and testable

An agent is not “an SAP expert” by assertion. Its required methodology, technical, business-process, industry and localization skills should be named, versioned and evaluated.

Evidence precedes material action

Important recommendations should identify their sources, assumptions and confidence. Important actions should require attributable authority.

The mesh must be operable

Every production agent needs ownership, monitoring, release management, incident handling, cost management and a path to retirement. A successful pilot is not yet an enterprise service.