Dify

Dify is an open-source LLM application development platform that provides a visual interface for building AI agents, chatbots, RAG pipelines, and workflow automations. With 55,000+ GitHub stars, it has rapidly become one of the most popular tools for teams wanting to build production AI applications without deep infrastructure work.

The platform's visual workflow editor lets users design complex AI pipelines by connecting nodes for LLM calls, knowledge retrieval, code execution, conditional logic, HTTP requests, and tool use. It supports agent mode where LLMs autonomously select and use tools based on the task, as well as structured workflow mode for deterministic pipelines.

Dify's knowledge base system handles document ingestion, chunking, embedding, and retrieval with support for multiple vector databases (Weaviate, Qdrant, Pinecone, pgvector). Users can upload PDFs, Word docs, web pages, and structured data, then connect this knowledge to any application for grounded responses.

The platform supports all major LLM providers (OpenAI, Anthropic, Google, Azure, local models via Ollama) with unified API access and model management. It includes prompt engineering tools, annotation-based fine-tuning, built-in analytics, and API publishing for integrating AI features into existing products.

Dify can be self-hosted via Docker or used through its cloud service. The open-source edition includes most features, making it accessible for startups and individual developers while enterprise features cover SSO, audit logs, and custom model hosting.