RAG (ingest)

RAG (ingest) is the “write” half of RAG: it takes text, splits it into chunks, computes their embeddings and stores them in the vector store under a namespace. Later, RAG (retrieve) searches over that content.

It’s a flow node (it does not connect to the agent). You use it in an ingestion workflow that you run once, or whenever your documents change.

Configuration

Field	Type	Required	Description
namespace	text	Yes	Namespace to store into.
provider	select	Yes	Embeddings provider.
model	text	Yes	Embeddings model (the same you’ll use when retrieving).
credentials_id	credentials	Yes	Provider credential.
chunk_size	number	No	Size of each chunk in characters. Default 1000.
overlap	number	No	Overlap between chunks. Default 150.

Input: data.text (a single text) or data.documents (an array [{ text, id, metadata }]). Output: data.indexed (how many chunks were stored) and data.namespace.

Example — Index a manual

1. Start → (a node that brings the text: read a file, a table, an HTTP call…) → RAG (ingest) → End.
2. RAG (ingest):
   • namespace: product-manual
   • provider: openai, model: text-embedding-3-small
   • credentials_id: your credential
   • chunk_size: 1000, overlap: 150
3. Run the workflow once. Confirm the output has indexed > 0.

From then on, an agent with RAG (retrieve) pointing at the same namespace and same model can answer about that manual.

Caution

The indexing embeddings model must be identical to the retrieval one. If you index with text-embedding-3-small and retrieve with another, the vectors aren’t comparable and the search fails.

More specific chunks

If answers mix topics, lower chunk_size (smaller, more precise chunks) and/or raise top_k when retrieving.

See also

• RAG (retrieve) · AI Agent (orchestrator)
• Examples and recipes · Common problems