Text Embedding

Generate vectors for document text and attach them to the original document records.

Use this node when later workflow steps need documents with embeddings. To build a searchable knowledge base directly, use Embed and Add to Knowledge Base.

Configure input and model

Input or setting

Description

documents

Required nonempty array, usually from parsing, cleaning, or chunking

model

Text embedding model; defaults to the workspace text embedding service configuration

Choose a model that supports text embeddings. If no model is selected, the workspace must have an available text embedding service configured.

Use the output

The output is still documents. The node embeds nonempty content into each record’s embedding field and retains metadata.

Vectors remain in the output documents. Connect a writing node to persist them, or a knowledge base writing node to build a retrieval index.

Example: embed chunks

Flow: General Document Parsing → Segment → Text Embedding.

Pass the full chunk array as documents and select a text embedding model. Check that each record contains both its text and the corresponding vector.

Last updated on