Principle:FlowiseAI Flowise Document Store Creation

Attribute	Value
Sources	packages/ui/src/api/documentstore.js
Domains	Document_Store_Ingestion
Last Updated	2026-02-12 14:00 GMT

Overview

Document_Store_Creation is a technique for creating named document store containers that organize ingested documents for vector search retrieval. A document store is the foundational organizational unit in the FlowiseAI document ingestion pipeline, providing a logical boundary for grouping related documents, their chunked representations, and vector embeddings.

Description

A document store is a logical container that groups related documents, their chunked representations, and vector embeddings. Creating a store is the first step in the RAG (Retrieval-Augmented Generation) pipeline. Each store has a name and description and serves as the organizational unit for document management.

The creation process involves:

• Naming the store -- Each store requires a unique, descriptive name that identifies the document collection it contains.
• Adding a description -- An optional description provides context about the purpose and contents of the store.
• Persisting to the backend -- The store configuration is sent to the server via a POST request and persisted to the database, returning a unique identifier.

Once created, the document store becomes available for:

• Adding document loaders that ingest content from various sources
• Configuring text splitters for chunking
• Previewing and editing chunks
• Upserting embeddings into vector stores
• Querying for similarity search retrieval

Usage

Use document store creation when setting up a new document collection for RAG-based chatflow retrieval. Typical scenarios include:

• New knowledge base -- Creating a store to hold product documentation, FAQs, or support articles for a customer service chatbot.
• Domain-specific retrieval -- Isolating legal documents, medical records, or technical specifications into separate stores for targeted retrieval.
• Multi-tenant isolation -- Creating separate stores per tenant or project to ensure document isolation.

// Creating a new document store via the API
const response = await documentStoreApi.createDocumentStore({
    name: 'Product Documentation',
    description: 'Technical docs for product v2.0'
})
// response.data contains the created store with its unique id

Theoretical Basis

Document store creation follows a container-based organization pattern for document management. This pattern provides several advantages:

• Isolation -- Each store provides complete isolation between different document collections. This means separate embedding configurations, vector store backends, and retrieval parameters can be applied per collection without interference.
• Lifecycle management -- Stores serve as the unit of lifecycle management. Documents within a store share the same processing pipeline (loader, splitter, embedder, vector store), making it straightforward to re-index or delete an entire collection.
• Composability -- In the broader RAG architecture, stores can be composed as tools or retrieval sources within chatflows, enabling flexible multi-source retrieval strategies.
• Organizational semantics -- By grouping documents into named stores, the system provides a human-readable organizational layer on top of the underlying vector indices, making it easier to manage large numbers of documents.

This pattern is analogous to database schemas or S3 buckets -- providing logical grouping with independent configuration and access control.

Related Pages

• Implementation:FlowiseAI_Flowise_CreateDocumentStore
• Principle:FlowiseAI_Flowise_Document_Loader_Selection -- Next step: adding document loaders to the store
• Principle:FlowiseAI_Flowise_Chunk_Preview -- Previewing chunks after loader configuration
• Principle:FlowiseAI_Flowise_Vector_Store_Upsert -- Upserting processed chunks into vector stores