Phase 6: Dify & Visual AI App Builders · 50 min · Dify · Docker
Dify Platform Overview — Visual AI App Building
Build AI apps visually, deploy them as APIs.
Hiring signal: Dify is the leading open-source AI app builder. Being able to explain Dify's positioning (visual AI builder vs n8n's workflow automation), its app types (chatbot, agent, text generator, workflow), and its self-hosting model demonstrates broad platform awareness that automation consultant and solutions architect roles require.
What you will learn
- Understand Dify's positioning: visual AI app builder optimized for LLM applications
- Compare Dify app types: Chatbot, Agent, Text Generator, Workflow
- Self-host Dify with Docker Compose for data privacy and unlimited usage
- Understand Dify's model provider integration: OpenAI, Anthropic, Ollama, local models
The Problem
You need to build an AI chatbot that answers questions from company documents. You could write a LangChain app with Python, set up a vector store, build a web interface, and deploy it. That's 500+ lines of code and days of work. Or you could use Dify: upload documents to a knowledge base, create a chatbot app, configure the model, and publish it as an API — all through a visual interface, no code required.
Dify is a visual AI app builder. It's not a workflow automation platform like n8n — it's specifically designed for building LLM-powered applications (chatbots, agents, text generators, workflows) with a drag-and-drop interface.
Dify builds AI apps; n8n automates workflows
Dify is optimized for AI: knowledge bases, model configuration, prompt engineering, agent tools. n8n is optimized for integration: 400+ app connectors, webhooks, scheduling, data transformation. Use Dify when the core task is AI processing. Use n8n when the core task is connecting systems. Use both together for production AI automation.
The Concept
Dify vs n8n vs LangChain
| Aspect | Dify | n8n | LangChain |
|---|
| Primary purpose | Visual AI app builder | Workflow automation | AI programming framework |
| Interface | Visual (drag-and-drop) | Visual (node-based) | Code (Python/JS) |
| AI features | Knowledge base, agents, prompts | AI Agent nodes | Full control |
| Deployment | API, web app, widget | Webhook, scheduled | Custom |
| Self-hosting | Docker Compose | Docker | N/A (library) |
| Learning curve | Low (visual) | Low (visual) | High (code) |
| Flexibility | Medium | Medium | High |
| Best for | AI apps, chatbots, RAG | Integrations, automation | Custom AI systems |
Dify App Types
┌──────────────────────────────────────────────────────────────┐
│ DIFY APP TYPES │
│ │
│ 1. CHATBOT │
│ ├── Conversational AI with memory │
│ ├── Connect to knowledge base for RAG │
│ ├── Deploy as: web widget, API, embed │
│ └── Best for: customer support, Q&A, assistant │
│ │
│ 2. AGENT │
│ ├── Autonomous AI with tools │
│ ├── Strategies: Function Calling or ReAct │
│ ├── Tools: built-in, custom HTTP, workflows, knowledge base │
│ └── Best for: dynamic tasks requiring tool use │
│ │
│ 3. TEXT GENERATOR │
│ ├── Single input → single output (no conversation) │
│ ├── Templates with variables │
│ ├── Deploy as: API │
│ └── Best for: summarization, translation, classification │
│ │
│ 4. WORKFLOW │
│ ├── Multi-step visual pipeline │
│ ├── Nodes: LLM, knowledge retrieval, code, HTTP, branching │
│ ├── Deploy as: API │
│ └── Best for: complex AI pipelines with branching │
└──────────────────────────────────────────────────────────────┘
You need to build a system that: receives emails, classifies them, extracts data, and creates records in a CRM. Which Dify app type is best?
Workflow is best because the task is a multi-step pipeline: classify → extract → create record. Each step has a specific purpose and the path may branch (different CRM actions based on classification). Dify workflows support LLM nodes, code nodes, HTTP request nodes, and conditional branching — exactly what this pipeline needs. An agent would be overkill (the path is mostly known). A chatbot is conversational, not pipeline-oriented. A text generator is single input/output.
Dify Model Provider Integration
| Provider | Models | Setup |
|---|
| OpenAI | GPT-4o, GPT-4o-mini, o1 | API key |
| Anthropic | Claude 3.5 Sonnet, Haiku | API key |
| Ollama | Llama, Mistral, Phi (local) | Ollama URL |
| Local models | Any via Ollama/vLLM | Local server URL |
| Azure OpenAI | GPT models on Azure | Azure endpoint + key |
Unlock the full lesson
You've read the first 2 sections. The rest of this lesson covers Build It, Use It, Ship It, Exercises, Key Terms, Common Pitfalls — plus a hands-on lab, quiz, and project artifact.
Create a free account to unlock Phase 0 and Phase 1 of every course — no credit card.
Browse all courses · View pricing · DeVenture Academy