RAG Pipeline

Multi-format ingestion, semantic chunking, and dual vector + graph retrieval for accurate AI responses

manage_search

Overview

The OpenRails RAG (Retrieval-Augmented Generation) Pipeline is the intelligence backbone of the platform. It transforms raw documents into searchable, contextual knowledge that powers AI chat responses, agent workflows, and automated analysis. Unlike basic vector-only RAG implementations, OpenRails uses a dual retrieval architecture combining vector search with knowledge graph RAG for relationship-aware context retrieval.

Key Differentiator: Dual retrieval means OpenRails understands not just semantic similarity (what things are about) but also entity relationships (how things connect). This produces significantly more accurate and complete answers than vector-only systems.

stars

Key Capabilities

upload_file Multi-Format Ingestion

Process PDF, DOCX, PPTX, TXT, HTML, Markdown, images (via OCR), video (frame extraction + transcription), and audio (speech-to-text transcription). Each format has optimized extraction logic.

content_cut Semantic Chunking

Intelligent document splitting that respects paragraph boundaries, section headers, and semantic coherence. Configurable chunk sizes with overlap for context continuity across chunk boundaries.

scatter_plot Vector Search

High-performance embedding storage and similarity search. Supports multiple embedding models, configurable distance metrics, and filtered search by metadata, collection, or security level.

share Graph RAG

Entity and relationship extraction builds a knowledge graph. Queries traverse entity connections to find contextually relevant information that vector search alone would miss.

folder_special Data Lakes & Collections

Organize ingested content into collections (data lakes) for scoped retrieval. Control which knowledge bases are available to which projects, users, or agents.

auto_fix_high Auto-Embedding

Documents are automatically chunked and embedded upon ingestion. Background processing handles large batches without blocking the UI. Progress tracking for bulk uploads.

grid_view

Chunking Strategies

StrategyBest ForChunk SizeOverlap
SemanticLong-form documents, reports, manualsConfigurableAdjustable based on content type
ParagraphWell-structured documents with clear sectionsVariableAdjustable based on content type
Fixed-SizeUnstructured text, logs, transcriptsConfigurableConfigurable
RecursiveNested documents, code filesConfigurableConfigurable
Best Practice: Chunking settings can be tuned per data lake to optimize retrieval quality for your specific content types.
compare

Dual Retrieval Architecture

Query Flow:

User QueryEmbedding Generation → parallel: [ Vector Search | Knowledge Graph Traversal ] → Result FusionRanked ContextLLM Prompt

scatter_plot Vector Path

Finds chunks semantically similar to the query. Excels at "what is X?" and "tell me about Y" questions where topical relevance is key.

share Graph Path

Traverses entity relationships to find connected information. Excels at "how does X relate to Y?" and "what are the dependencies of Z?" questions.

settings

Technical Details

ComponentTechnologyConfiguration
Vector StoreVector DatabaseOptimized similarity matching with filtering
Graph StoreKnowledge Graph DatabaseEntity extraction, relationship mapping, graph queries
Embedding ModelsConfigurableOpenAI, local self-hosted models, custom models
Chunk StorageRelational Database + Object StorageMetadata, lineage tracking, version history
Processing QueueHigh-performance processing pipelineAsync ingestion, retry logic, progress tracking
lightbulb

Use Cases

library_books

Knowledge Base

Turn thousands of documents into a searchable AI-powered knowledge base with cited answers

policy

Compliance Research

Query regulatory documents and internal policies with relationship-aware retrieval

biotech

Research Synthesis

Ingest research papers and find connections across studies using graph traversal

Related Feature Sheets