Docling Pipelines

A modular, operator-based data processing framework for building flexible document curation pipelines and Retrieval-Augmented Generation (RAG) workflows.

Python 3.12 21 Operators IBM Open Source RAG Ready Docling Powered

Getting started

Ready to kick off your Docling Pipelines journey? Start here:

Documentation Book

Want the full narrative guide? The 12-chapter book covers every topic end-to-end — architecture, flow authoring, all operators, LLM integration, production deployment, and contributing.

Docling Pipelines — The Complete Guide 12 chapters · AsciiDoc · Built with Antora Architecture → Installation → Flows → Operators → Quality → Vector Storage → LLM → Python API → Extending → Production → Troubleshooting
Open the book →

Visual UI Guide

Prefer a visual, drag-and-drop experience? The UI Guide walks through building a full multi-source RAG pipeline on the Docling Pipelines canvas — from ingestion to live observability — with annotated screenshots of every operator panel.

Building Multi-Source RAG with the Docling Pipelines UI Visual canvas · Operator walkthrough · Live observability Ingest → Extract → Quality Guards → Hybrid Chunking → VectorDB Mapping → Live Run Monitoring
Open the UI Guide →

I want to…

Get started quicklyQuick Start Guide
Set up my environmentSetup Guide
Write my first flowFlow Configuration Guide
Look up a parameterGlobal Config  ·  Operator Reference
Use the Python APIPython API Guide
Use the REST APIREST API Server
Create a custom operatorCustom Operators Guide
Connect a new data sourceCreate Connector Guide
Secure my deploymentSecurity Best Practices
Deploy to productionAdvanced Configuration
Read the full narrative guideDocumentation Book
Build visually with the UIVisual UI Guide
Troubleshoot an issueTroubleshooting Guide
Contribute codeContributing Guide