Langflow FAQ v1.4.0 (LTS)

How do I improve retrieval quality in a RAG flow?

Poor retrieval results are usually caused by one of the following:

• Wrong embedding model — the model used to create the knowledge base embeddings must match the model used at query time. Check the model_name used in aidb.create_table_knowledge_base.
• Chunk size — if source documents were ingested as large chunks, retrieval scores may be noisy. Re-index with smaller chunks using aidb.chunk_text_config.
• Top-k too low — the EDB Knowledge Base component returns the top-k most similar results. Increase this if relevant content is being missed.
• Index not up to date — if source data has changed since the last aidb.bulk_embedding run, retrieved content may be stale. Re-run the bulk embedding or set up incremental syncing.

What's the difference between EDB Model Server and EDB Embeddings?

EDB Model Server sends a prompt to an LLM and returns a text response — use it for generation, summarisation, and reasoning steps. EDB Embeddings converts text into a vector using an embedding model — use it when you need to compute similarity or store vectors in Postgres. Both require a model cluster to be deployed in your HM project, but they use different model types (LLM vs embedding model).

"Can't select a project" — the portal domain defaults to the API endpoint

Select Controls at the top of the affected EDB Database or Knowledge Base component and set Hybrid Manager URL to:

http://istio-internalgateway.istio-system.svc.cluster.local:7080