AI Stack Map
A map of roles in an AI application. Each role links to directories, rankings, guides, and labs that already exist.
Open one role to see the options for that role. This page does not score tools or combine tools from different roles.
Runtime
Application roles
Capabilities a running AI application can include. Follow a role to the pages that already cover it.
Models
Foundation models such as GPT, Claude, Gemini, and Llama. Model pages and the general-purpose ranking sit beside the tool directory.
Retrieval
Semantic search and embedding storage platforms.
RAG frameworks
Frameworks and platforms for retrieval-augmented generation. These are application frameworks, separate from the vector database directory.
Agents
Frameworks and platforms for autonomous AI agents.
Evaluation
Evaluate prompts, models, and RAG systems.
Observability
Monitor, debug, and improve AI applications.
Serving
Deploy, optimize, and scale AI inference.
Gateway and routing
A gateway is the boundary for accessing model providers. Routing chooses a model or provider for each request.
Build
Coding tools
Coding tools help you build the application. They are separate from the runtime roles above, which are the parts of a running application.
Coding tools
Code assistants, IDEs, and developer productivity tools for writing software.
Adjacent
Adjacent capabilities
Knowledge graphs, guardrails, and seed landscapes sit beside the runtime roles. Landscape pages are seed overlays: a small set of entity records and the relationships stored on those records.
Knowledge graphs
Structured entities and relationships for GraphRAG and enterprise search.
Guardrails
Input and output validation for safe LLM applications.
Seed landscapes
Seed overlays from entity records already on the platform. These pages show that small graph. They are not a map of the full tool directory.