Skip to content

AI Addon Backend

Related documentation: Addon Modules Backend · Chat + MCP (Node) · Backend MCP (Go) · RAG KB LLM · AI Event Intelligence Copilot Prototype

The ai addon exists in the module taxonomy and already has multiple real runtime repos in this workspace.

  • implemented
  • Node/TypeScript AI agent stack
  • LangChain/Fastify agent + MCP tool microservices + optional Postgres chat memory
  • adjunct tooling, not Auth/Core/Finance/master-data ownership
  • implemented
  • Cloudflare Workers MCP server over WebSocket
  • lightweight MCP tool surface for compatible clients such as n8n
  • implemented
  • RAG ingestion + retrieval service using PostgreSQL source data, OpenAI embeddings, and Qdrant
  • supports retrieval/chat use cases for artist/market directory knowledge
  • implemented as a separate prototype repo
  • adjunct AI product/prototype, not platform master-data ownership

What is not implemented as one unified service

Section titled “What is not implemented as one unified service”

What does not exist yet is a single consolidated production Backend-Kisum-AI service that replaces all of the above.

So docs should say:

  • AI addon capability is real
  • the runtime is split across multiple repos
  • consolidation is still TODO

Do not describe the AI addon as “not published yet” or “future only.” That is stale.