Dify#

Dify is an open-source LLM application development platform with a visual orchestration interface that lets developers build AI applications without writing much code.


What It Can Do#

1. Build AI Apps Visually#

  • Chat assistants: configure conversation flows via drag-and-drop, with support for multiple LLMs (OpenAI, Claude, local models)
  • Workflow orchestration: a visual DAG editor that chains LLM, code, HTTP, conditional-branch, and other nodes
  • Agents: configure tool calling so the LLM can plan and execute tasks autonomously
graph LR
    A[User input] --> B[Intent detection]
    B --> C{Branch}
    C -->|Query| D[Knowledge base retrieval]
    C -->|Action| E[Tool call]
    D --> F[LLM generation]
    E --> F
    F --> G[Output]

2. Knowledge Base / RAG#

  • Upload documents (PDF, Word, Markdown, web pages)
  • Automatic chunking, embedding, and storage
  • Relevant content is retrieved automatically during conversations to improve answer accuracy

3. Multi-Model Management#

Supported ModelsExamples
Commercial APIsOpenAI, Claude, Gemini, Tongyi Qianwen
Open-source modelsLLaMA, Qwen, ChatGLM (via Ollama/vLLM)
Embeddingtext-embedding-3, bge, m3e
RerankCohere, bge-reranker

4. One-Click Deployment#

  • Generates a WebApp once built (shareable link)
  • Provides an API (REST, can be integrated into your own apps)
  • Can be embedded in web pages (iframe / JS SDK)

5. Other Capabilities#

  • Prompt template management: version control, A/B testing
  • Conversation logs and annotation: review user conversations and annotate them for fine-tuning
  • Team collaboration: share apps and knowledge bases across a team
  • Self-hosting: one-command Docker deployment, data stays on-premises

Use Cases#

ScenarioDescription
Enterprise knowledge base Q&AUpload internal documents for employee self-service lookup
Customer service botConnect product docs to answer user questions automatically
Internal toolsOrchestrate workflows to automate tasks
AI app prototypingValidate ideas quickly without building from scratch

Quick Start#

# Docker deployment
git clone https://github.com/langgenius/dify.git
cd dify/docker
docker compose up -d

# Visit http://localhost/install to complete initialization

Website: dify.ai | Repository: github.com/langgenius/dify