Reference Architecture

A high-level view of how AI services fit together.

Agentic Friendly

AI combines product-facing interfaces with tenant-aware application services, model serving, retrieval, tools, and observability.

Architecture view

AI domains

DomainMain roleExample components
User experienceProvide chat, files, Agents, MCP, models, and administrationAI Web Portal
API and orchestrationExpose tenant-aware application services and workflowsBSQAI API, Temporal
Tenant lifecycleReconcile tenant identity and optional platform resourcesBSQAI Tenant Operator, PlatformTenant
Model lifecycleImport, register, and observe modelsModel Installer, MLflow
Model access and inferenceRoute and execute model requestsLiteLLM, KServe, vLLM
Tools and searchConnect approved MCP and web-search providersMCP Gateway, MCP Lifecycle Operator
Retrieval and documentsConvert, embed, store, and retrieve contentDocling, Milvus
AI observabilityTrace application and inference operationsOpenTelemetry, MLflow, Grafana Alloy

Design intent

Applications integrate with the portal or BSQAI API. Internal model, storage, workflow, and tool components can evolve without changing those primary product interfaces. Tenant identity and authorization remain attached as requests move through each layer.

On this page