Dify

Getting Started

1. Self-host Dify using Docker Compose: docker compose up -d from the cloned repository, or sign up for Dify Cloud.
2. Configure your LLM provider API keys in the settings panel to connect Claude, GPT-4, or other models.
3. Create a new app using the visual workflow builder, choosing from chatbot, text generator, or agent templates.
4. Add knowledge bases for RAG by uploading documents, and publish your app with a shareable URL or API endpoint.

Key Features

• Visual workflow builder enables drag-and-drop construction of complex LLM pipelines with branching and iteration.
• Built-in RAG engine with document ingestion, chunking, embedding, and retrieval for knowledge-grounded responses.
• Multi-model support connects to hundreds of LLM providers and lets you switch models without changing app logic.
• Agent mode builds autonomous agents with tool access, code execution, and web browsing capabilities.
• One-click deployment publishes apps as web interfaces, chat widgets, or API endpoints instantly.
• Self-hostable with full Docker support and enterprise features for data privacy and access control.