Unified AI and MCP

Use one governed Hub AI service and Model Context Protocol path so assistants and agents work consistently across platform products.

OL Hub unifies AI enablement and MCP (Model Context Protocol) so design assistants, planners, and external agents share authentication, configuration conventions, and transport patterns across SaaS and OnPrem.

Capabilities

Hub AI service

Python ai-service for contract generation, auto-mapping, expressions, field descriptions, Planner Assistant chat, modernization streams
Hub Enterprise: AI as a separate container (hub-ai-service) next to HE; toggle via OL_AI_ENABLED_KEY
Light Hub: AI can run embedded as a subprocess for demos
SaaS: AI microservices in the managed mesh; auth via Hub API keys / JWT

Hub MCP server

Backend MCP over SSE (GET /api/v1/mcp/sse) and Streamable HTTP (POST /api/v1/mcp)
JWT / API-key authentication aligned with Hub security
Tools, resources, and prompts (e.g. project info, documentation, hub-assistant)