Capability Domains
The main capability areas covered by CoreAI.
CoreAI is easier to understand when viewed as a set of capability domains rather than a list of unrelated products.
User Interface And API For GenAI And Agents
This domain provides the main entry points into the platform.
| Capability | Main components | Purpose |
|---|---|---|
| Web experience | CoreAI Web Portal | Main user surface for AI workflows |
| Documentation | Documentation | Product guidance and reference material |
| Backend services | CoreAI API, Model Installer | Application logic, orchestration, and model delivery |
GenAI And ML Lifecycle
This domain manages how models are tracked, versioned, and prepared for use.
| Capability | Main components | Purpose |
|---|---|---|
| Model registry | MLflow | Track experiments and manage model versions |
| Model delivery | Model Installer | Install and prepare models for Runtime inference |
LLM Proxy And MCP
This domain standardizes access to models and tools.
| Capability | Main components | Purpose |
|---|---|---|
| LLM gateway | LiteLLM | Governed access to model providers and endpoints |
| MCP operations | ToolHive | Deploy and manage MCP server integrations |
RAG And Vectors
This domain supports retrieval and document understanding use cases.
| Capability | Main components | Purpose |
|---|---|---|
| Document parsing | Docling | Convert documents into structured content |
| Vector search | Milvus | Store and query embeddings for retrieval workflows |
Logs And Observability
This domain gives teams visibility into AI behavior and platform signals.
| Capability | Main components | Purpose |
|---|---|---|
| GenAI observability | OpenTelemetry, MLflow | Trace prompts, responses, tool calls, and application behavior |
| Telemetry standard | OpenTelemetry | Export service telemetry in a consistent format |
Why This Matters
This separation helps teams understand where a capability belongs, what it depends on, and how it fits with Runtime and the higher-level product layers.