langgenius/dify

Dify is an open-source platform for building applications powered by large language models. Its short description calls it a production-ready platform for agentic workflow development. In plain terms, large language models are the AI brains behind chatbots and assistants, and Dify gives you a visual interface to wire them up into real apps without writing all the plumbing from scratch. An agent is software that can decide which steps to take, and a workflow is a defined sequence of steps, Dify is built for either approach. The README lists key features. There is a visual workflow canvas where you build and test AI workflows. Comprehensive model support means Dify integrates with hundreds of LLMs from dozens of providers, covering GPT, Mistral, Llama3 and anything that speaks the OpenAI API. A Prompt IDE lets you craft prompts, compare model performance, and add extras like text-to-speech. The RAG pipeline, short for retrieval-augmented generation, covers document ingestion through retrieval with out-of-box text extraction from PDFs, PPTs and other common document formats. Agent capabilities let you define agents using LLM Function Calling or the ReAct pattern, with pre-built or custom tools. Observability integrations are mentioned for Opik, Langfuse and Arize Phoenix. There is a hosted Dify Cloud as well as self-hosting, the Quick Start uses Docker Compose with at least two CPU cores and 4 GiB of RAM. You would actually use Dify if you want to build a chatbot, internal AI tool, RAG-based knowledge assistant or an autonomous agent for your team or product. The repo's primary language is TypeScript, and the standard deploy is Docker and Docker Compose.