Skip to main content

The AI Periodic Table: Your Mental Model

Just as chemistry's periodic table organizes elements into families with predictable properties, the AI Periodic Table organizes AI concepts into families and complexity levels. Once you understand this structure, you can decode any AI architecture, any product demo, any technical pitch.

Credit & Foundation

The AI Periodic Table concept was presented by Martin Keen from IBM Technology. We've adapted and extended their framework to create Learn AI at Jahnel Group.

Watch: AI Periodic Table Explained by IBM →


The Table

Rows = Complexity Levels

From atomic primitives to cutting-edge emerging technologies:

RowNameDescription
Row 1PrimitivesThe atomic building blocks that can't be broken down further
Row 2CompositionsCombinations of primitives that form practical tools
Row 3DeploymentProduction-ready systems and practices
Row 4EmergingCutting-edge developments still rapidly evolving

Groups (Columns) = Functional Families

Concepts that serve similar purposes at different complexity levels:

GroupFamilyTheme
G1ReactiveAction—from instructions to autonomous operation
G2RetrievalMemory—storing, finding, and adapting information
G3OrchestrationCoordination—combining components into systems
G4ValidationQuality—ensuring systems work correctly and safely
G5ModelsIntelligence—the raw capabilities that power everything

How to Use This Table

  1. Learn the families first. Understanding the five groups helps you categorize any new AI concept you encounter.

  2. Follow the rows for complexity. Row 1 concepts are prerequisites for Row 2. Row 2 enables Row 3. And so on.

  3. Notice the connections. RAG (Orchestration) depends on Embeddings and Vector DBs (Retrieval). Agents (Reactive) often use Function Calling (Reactive) and Frameworks (Orchestration).

  4. Map real systems. When you see an AI product or architecture, identify which elements are in play. It demystifies the complexity.