Dify

Open-source platform for building LLM apps with visual workflows and RAG

AI Agents & AutomationFree planOverseasβ˜…β˜…β˜…β˜…β˜† 4.0

What is Dify?

Dify is an open-source development platform for LLM applications that combines a visual workflow builder, retrieval-augmented generation, agent tooling and prompt management in one interface. You connect a model provider, upload knowledge documents, and assemble a chat assistant, agent or automation pipeline without writing the orchestration layer yourself. Because it self-hosts, teams keep control of their data and prompts, and the same project can be published as a web app or consumed through an API. It is a middle path between raw code and closed no-code products.

Last updated: 2026-09-20. This site only provides an index; for exact features, pricing, and licensing, see the official website.

Key features

  • Visual workflow builder for chatflows, agents and pipelines
  • Built-in RAG with document ingestion and vector retrieval
  • Prompt management and versioning across applications
  • Model-agnostic provider configuration, including local models
  • Publish as a hosted web app or consume via API
  • Self-hosted deployment with Docker or Kubernetes

Pros & cons

Strengths

  • One platform covers RAG, agents and prompt operations
  • Self-hosting keeps data and prompts under your control
  • Open source avoids lock-in to a single vendor

Watch out for

  • Self-hosting carries upgrade and maintenance work
  • Visual workflows get unwieldy for very complex logic
  • Retrieval quality still depends on how you prepare documents

Best for & use cases

internal tools, knowledge assistants, rag applications and llm prototypes

If you're comparing similar products, check the alternatives below, or browse all tools in the AI Agents & Automation category.

FAQ

What can I build with Dify?

Chat assistants, agents and workflow automations over your own documents. Typical projects are internal knowledge bots, support assistants and pipelines that combine retrieval with tool calls.

Do I have to host it myself?

No. There is a managed cloud option. Self-hosting is popular because it keeps documents and prompts inside your own infrastructure, at the cost of running upgrades yourself.

How good is the retrieval?

It provides the pipeline, but quality depends on your documents: clean, chunked, well-labelled sources retrieve far better than a dump of mixed files with no structure.