Tutorial Progress
Step 1 of 4

Getting Started with Dify

beginner 15 minutes Fundamentals

Learn the basics of Dify AI platform, how to access it, and understand core concepts like LLMs, embeddings, and context windows.

1 What is Dify?

3 min

Dify is an open-source, user-friendly LLMOps (Large Language Model Operations) platform designed to simplify and accelerate the development, deployment, and management of AI applications.

Key Features:

  • Visual AI Orchestration: Build AI applications using a drag-and-drop interface
  • No-Code/Low-Code: Perfect for beginners with no programming experience
  • RAG Engine: Built-in Retrieval-Augmented Generation for knowledge-based applications
  • AI Agent Framework: Create intelligent agents that can use tools and make decisions
  • Model Support: Works with all mainstream LLMs including GPT, Claude, and open-source models

Why Choose Dify?

Unlike other AI tools that offer individual components, Dify provides a comprehensive, production-ready solution. Think of it as a well-designed scaffolding system that handles the complexity while you focus on creating innovative AI applications.

Tips:
  • Dify is completely open-source and free to use
  • You can use either the cloud version at dify.ai or deploy locally
  • No coding experience required to get started

2 Accessing Dify

4 min

There are two ways to access Dify:

Cloud Version (Recommended for Beginners)

  1. Go to dify.ai
  2. Sign up for a free account
  3. Verify your email address
  4. Start building immediately

The cloud version includes free usage quotas for popular AI models, making it perfect for learning and experimentation.

Local Installation (For Advanced Users)

  1. Clone the repository from GitHub
  2. Install Docker and Docker Compose
  3. Run docker compose up -d in the dify/docker directory
  4. Access at http://localhost/install
Tips:
  • Start with the cloud version for easier setup
  • Local installation gives you full control over data
  • Both versions have the same features and capabilities

3 Understanding Core Concepts

5 min

Before building your first AI application, it's important to understand these key concepts:

Large Language Models (LLMs)

LLMs are AI models trained on vast amounts of text data. They can understand and generate human-like text. Popular LLMs include:

  • OpenAI GPT-4: Excellent for complex reasoning and creative tasks
  • Claude: Great for analysis and following instructions
  • Open-source models: Cost-effective alternatives for specific use cases

Context Window

The context window is the amount of text an LLM can "see" and "remember" at once. A larger context window allows the model to consider more information when generating responses, leading to more accurate and coherent outputs.

Embeddings

Embeddings convert text into numerical vectors that capture semantic meaning. Similar texts have similar vectors, enabling AI to understand relationships between different pieces of content. This is crucial for search and retrieval functions.

Tokens

Tokens are the basic units that LLMs process - roughly equivalent to words or parts of words. Understanding tokens helps you manage costs and context limits.

Tips:
  • Different LLMs have different strengths - experiment to find the best fit
  • Larger context windows cost more but provide better results for complex tasks
  • Embeddings enable semantic search, not just keyword matching

4 Exploring the Dify Interface

3 min

Once you're logged into Dify, you'll see a clean, intuitive dashboard with several key sections:

Main Navigation

  • Studio: Where you build and manage your AI applications
  • Knowledge: Manage your knowledge bases and documents
  • Tools: Configure external tools and integrations
  • Datasets: Manage your training data and examples

Application Types

Dify supports several types of AI applications:

  • Chatflow: Conversational applications with complex workflows
  • Assistant: AI agents that can use tools and external services
  • Completion: Text generation and completion applications

Quick Actions

  • Create from Blank: Start with a clean slate
  • Use Template: Begin with pre-built examples
  • Import: Bring in existing configurations
Tips:
  • Spend time exploring the interface before building your first app
  • Templates are great for learning best practices
  • The preview feature lets you test applications before publishing