Skip to content
EN
English 简体中文 soon 日本語 soon

Dify

Agent and RAG application platform

Visit official site

What Dify is

Dify is an application platform for teams building with models. It covers autonomous agents and retrieval pipelines, and unlike a simple playground it spans the whole lifecycle: developing workflows, deploying them and managing them once they are live, with both visual and developer-oriented paths.

What you can do with it

  • Build retrieval pipelines over your own documents
  • Develop autonomous agent workflows
  • Deploy applications and keep them running
  • Manage versions and monitoring
  • Standardise how a team builds AI features

Who it is for

  • AI application development teams
  • Platform and engineering groups
  • Organisations moving beyond prototypes

What to watch out for

  • Self-hosting gives control but brings upgrades, backups and security patching onto your team
  • Agents connected to tools need permission boundaries and logging
  • Retrieval quality depends entirely on how your documents are prepared
  • Usage of third-party models through the platform is still governed by those providers' terms

Pros & cons

✓ What we like

  • Covers build through operation
  • RAG and agents in one platform
  • Self-hosting option available

! What to watch out for

  • Self-hosting carries maintenance burden
  • Retrieval quality depends on your data prep
  • Agent permissions need boundaries

FAQ

What is it mainly used for?

Developing, deploying and managing autonomous agents and retrieval pipelines.

Can it run on our own infrastructure?

A self-hosted path exists alongside the hosted offering.

What determines answer quality?

Mostly how well your source documents are prepared for retrieval.

Last reviewed: 2026-09-18

More LLM API platform tools

View all →

How we review