Dify
Also known as: dify.ai, dify llmops
Dify sits in the LLMOps space: tools designed to make the full lifecycle of building and operating LLM-powered apps more manageable. Its visual workflow editor lets teams design agent logic, RAG pipelines (retrieval-augmented generation, where the model looks up relevant documents before answering), and multi-step automations without writing orchestration code from scratch.
At 136,000+ GitHub stars, it is one of the top three most-starred AI agent frameworks. Its appeal is breadth: built-in support for multiple model providers (OpenAI, Anthropic, open-source models), usage monitoring, MCP integration, and both cloud and self-hosted deployment. Teams can build everything from enterprise chatbots to internal workflow tools in one tool.
The audience for Dify is slightly different from pure code-first frameworks. Domain experts, product teams, and small engineering teams without dedicated ML infrastructure use it to get AI-powered services running quickly. For production systems that need fine-grained control over agent state, teams often graduate to LangGraph, but Dify handles a large portion of real-world AI product needs without that complexity.