Description:

We need a supported, scalable way to export the Port knowledge graph (blueprints, entities, runbooks, and use cases) as embeddings-ready data to load into our own vector database (e.g., Postgres + pgvector).

While Port AI's live query-time context is useful, we want to run advanced, low-latency RAG grounded in an embedded representation of our entire catalog. We would welcome either:

A documented API/export mechanism that supports incremental/delta syncs so we can chunk and embed the graph into our own DB.

A managed contract where we can point Port to our own vector database endpoint, allowing Port to manage the graph sync while we manage the storage.