sap

SAP Fiori and Compositional Design System evidence profile

Type: System profile
Status: Active
Canonical for: Factual interpretation of preserved SAP Fiori, OpenUI5 and recovered compositional-system evidence
Depends on: corpus/sap/ and its documented access boundary

Evidence boundary

SAP’s public documentation was materially inaccessible to the automated local capture, so the corpus contains the OpenUI5 source snapshot plus an access record. Under the research exception, the missing evidence was recovered from the official pages only. Those official pages are treated as documented; the underlying compositional schemas are not present locally and must not be treated as publicly implemented code. Access boundary, OpenUI5 snapshot

Four-question relevance

• Intent: documented through functional role, supported user intent and contextual fit.
• Structure: documented through relationship logic and structured metadata; schema and examples are unavailable.
• Retrieval: reportedly used by Joule Work, but the matching/retrieval mechanism is unavailable.
• Constraint/evaluation: generation-time compliance, accessibility and validation are documented; rules and evaluation results are unavailable.

AI product design

• Documented: SAP’s AI and Joule guidance includes five generative-AI principles, trust through transparency/control/bias prevention, prompt guidance, safe and sustainable AI, progressive explainability, matching, ranking, recommendations and feedback loops. Recovered official evidence
• Documented: Mobile guidance separates in-app AI components/patterns from Joule as a conversational copilot and possible entry point to multi-agent systems. Users should review and validate generated information. Recovered official evidence
• Implemented (conventional foundation): OpenUI5 supplies the public controls, APIs, types, tests, samples and accessibility foundations beneath Fiori. The snapshot does not expose an equivalent public package for the Compositional Design System.
• Evidence gap: The corpus does not permit a reliable public inventory of every current in-app AI component, its API or Storybook-equivalent examples. Product guidance is stronger than implementation visibility.

Compositional Design System

• Documented: SAP identifies a specific failure: an AI can assemble valid components into a “Frankenstein UI” that is technically correct but experientially incoherent because the relationships and intent remain in designers’ heads.
• Documented: SAP says a compositional system adds three things: relationship logic (which components can appear together and how), structured metadata (functional role, supported user intent and contextual fit), and generation-time guardrails (compliance, accessibility and validation active during generation).
• Documented: Joule Work “Spaces” are described as dynamically composed workspaces driven by intent and context. SAP states the Compositional Design System is already used there, not merely on a roadmap.
• Documented: Fiori remains the reliable transactional foundation; the compositional layer extends rather than replaces it. SAP distinguishes static transactional interfaces from dynamically assembled cross-system experiences.
• Documented: Design work shifts from screens to experience models, components to composition logic, guidelines to enforceable guardrails, and static systems to feedback/learning loops.
• Inferred: SAP supplies the most complete public theory of machine-readable design judgment in the corpus, but not the inspectable schema required to validate how relationships, intent and constraints are encoded.

Agentic behavior and governance

• Documented: SAP connects composition to an Autonomous Enterprise where assistants and agents execute end-to-end processes across systems.
• Documented: Governance is expected at generation time, including compliance, accessibility and validation, because post-hoc review is insufficient for dynamically generated experiences.
• Gap: Public material is not specific enough about permission request UI, action-level approval, interruption, resumption, audit history, rollback or identity delegation. These remain product/platform concerns in the visible evidence.

Machine readability

Clearly available

OpenUI5 exposes control identity, APIs, events, states, source code and tests. Human-readable Fiori guidance exposes pattern intent.

Claimed but not publicly inspectable

Relationship logic, supported intent, contextual fit, valid combinations, generation constraints and learning signals. These are precisely the semantics other systems expose only weakly, but SAP’s representation and enforcement mechanism are not in the corpus.

Evidence assessment

SAP is the strongest source for the proposition that AI composition requires more than component discovery. It is also the source with the largest public verification gap: an official production claim and detailed architecture direction without corresponding public schemas/code. The correct classification is documented and reportedly deployed, not independently implemented in the preserved public corpus.