The model in numbers
Four architecture planes
The cognition plane as a cross-cutting architectural plane alongside the data, control and management planes — as per the peer-reviewed EA4AI paper (Springer LNBIP).
Presentation of the paper at the EA4AI workshop (CBI-EDOC 2026): slides on ResearchGate →
Cognition Plane
Where AI agents reason, interpret and decide — under explicit mandate. With agents, mandates, reasoning boundaries, orchestration, grounding and explainability.
Control Plane
Where governance is enforced at runtime. Identity & access management, authorization, policy-as-code, audit trails and API gateways.
Data Plane
Where knowledge and data are managed. Vector databases, knowledge graphs, feature stores, RAG indices and document storage.
Management Plane
Where AI systems are operated. Configuration, deployment, monitoring & observability, model registry, lifecycle and resourcing.
Explore the model
The full model can be explored freely in SemanticXL, in the AI Body of Knowledge workspace. Search for concepts, knowledge areas and roles and follow the relationships between book, model and regulation.
Download the model
Want to work with it yourself? Download the model and import it into your own architecture tool.
The full context
This architecture model is based on the AI-BOK book, which elaborates all principles, roles, patterns and rationale in detail. Find the model valuable for yourself or your organisation? A book licence gives you the full context and supports further development.
View the editions