A workspace to build RAG and agent apps visually
Dify is LangGenius' visual AI app platform: RAG, agent workflows and many model providers in one workspace. Self-host with Docker Compose. License is Apache-2.0 plus conditions (multi-tenant SaaS needs a commercial license).
Updated 2026-09-04. Same-day snapshot of the official repo.
How to Get Started with Dify
3 steps to self-host with Docker Compose.
Clone and env
Official docker/README: enter docker/, copy the env file. Needs Docker Compose 2.24+.
git clone https://github.com/langgenius/dify.git
cd dify/docker
cp .env.example .envStart
Run Compose from docker/. Official minimum: 2 CPU / 4 GiB RAM.
docker compose up -dCreate admin
Open the install page on local port 80, create the admin, then sign in. If you changed ports, check .env.
http://localhost/installPitfalls
Two-layer data boundary
App data can stay on your servers; model calls still usually leave for an external API. Audit that hop — self-host ≠ data never leaves.
Don't spider-web the canvas
Past a dozen nodes, visual branches cost more than code. Use the canvas for PoC; split apps or drop to code when it gets messy.
Extra terms: multi-tenant and logo
Reselling Dify as multi-tenant AI, or stripping the console logo, hits the extra LICENSE terms. Read them before you productize.
Beside n8n & Ollama
Based on official docs and licenses. Dify is the app layer, n8n the flow layer, Ollama the model layer. They stack.
| Tool | License | Surface | Positioning | Best for |
|---|---|---|---|---|
| Dify | Apache-2.0 + conditions | AI app workspace | RAG + agent workflows + many models | SME PoC, knowledge Q&A |
| n8n | Fair-code · SUL | Visual workflows | System integration + cron/webhooks | Repetitive business flows |
| Ollama | MIT (models separate) | Local model runtime | One command to run open models locally | Private / offline inference |
Official Directory
Official
- Official site (dify.ai)
Product overview, cloud and self-host.
- GitHub repository
Source. Apache-2.0 with additional conditions.
Docs & Install
- Documentation
Self-host, RAG, workflows and model providers.
- Docker Compose deploy
Official self-host install path.
- docker/README in repo
Compose, .env and upgrade notes.
Related
- GitHub Releases
Read the target-version notes before upgrading.
Latest Updates
2026-09-04 · GitHub · langgenius/dify
Dify keeps shipping: agentic workflows + RAG in one workspaceFrom LangGenius; cloud, VPC or self-host. LICENSE is Apache-2.0 with extra conditions. 150k+ stars.
LaoJin's Pick
FAQ
- What is Dify?
- Dify is a visual AI app builder: RAG, agentic workflows and many model providers in one workspace. Self-host or use official cloud. Made by LangGenius.
- What should I know about Dify's license?
- Apache-2.0 plus conditions: operating a multi-tenant service (one workspace = one tenant) needs a commercial license; don't strip logo/copyright from the official frontend. Single-tenant internal use is typically fine. The repo LICENSE wins.
- How do I self-host Dify?
- Official docs want Docker Compose 2.24+ and at least 2 CPU / 4 GiB RAM. Clone, `cd dify/docker`, copy `.env.example` to `.env`, `docker compose up -d`. First visit the install page on local port 80 to create the admin.
- Dify vs n8n?
- Knowledge Q&A, chat apps, models as product features → Dify. Cron, webhooks, wiring existing SaaS/internal systems → n8n. Many stacks run both.
Need to deploy a team-level Agent system?
Open-source frameworks are free, but turning a runtime into an enterprise asset requires access control, RBAC sandboxes, and private RAG integrations. Laojin Global provides commercial-grade architecture and deployment support.
Cite This Hub
We encourage you to capture this hub into your knowledge base, research, or feed it straight to an AI agent under CC BY 4.0.
How this hub stays current
Version and links live in src/content/dify.ts; star snapshots sync via GitHub API.