How to Use Dify

Getting started with Dify

In this guide, you'll learn how to create AI agents and workflows using Dify's visual platform. By the end, you'll be able to set up a basic workflow and manage your AI models effectively.

Step 1: Sign up and set up

1. Go to dify.ai.
2. Click on the "Sign Up" button located in the top right corner.
3. Choose to sign up using your email or a third-party account (Google or GitHub).
4. Verify your email if prompted.
5. Once signed in, you’ll be taken to the dashboard where you can start building your workflows.

Step 2: Your first workflow

1. On the dashboard, click the "Create New Workflow" button.
2. Choose a template or start from scratch by selecting "Blank Workflow."
3. Drag and drop components from the left sidebar to the canvas. Start with a "Prompt" component.
4. Click on the "Prompt" component to configure it. Enter a simple question or command for your AI agent.
5. Add a "Response" component by dragging it onto the canvas. Connect the "Prompt" component to the "Response" component by clicking on the small dot on the right side of the prompt and dragging it to the response.
6. Click "Run" in the top right corner to execute your workflow and see the AI's response.

Step 3: Get better results

1. Experiment with different prompt styles to see how the AI responds. Use specific, clear instructions for best results.
2. Incorporate RAG (Retrieval-Augmented Generation) by adding a "Data Source" component. Connect it to your prompt to provide context.
3. Use the "Settings" menu to adjust model parameters, such as temperature and max tokens, to fine-tune outputs.
4. Save your workflows regularly by clicking the "Save" button located in the top right corner.

Pro tip

Take advantage of the "Templates" section to quickly set up common workflows. This can save you time and give you ideas on how to structure your own projects.

Common mistake to avoid

Avoid making prompts too vague. Specificity is key in guiding the AI to produce useful responses. Always test your prompts before integrating them into a more complex workflow.