AI Agent Workflow

1. Course Content

• Use multiple AI large models to build an agent workflow, implementing complex logical functions.

2. Start the Dify Service

• Connect to the vehicle via VNC or SSH, and enter the command in the terminal:

sh ~/bringup_dify.sh

• Check the IP address of the virtual machine by clicking "Robot Information" on the left side of the desktop.

3. Case: Classification Q&A Chatbot

• In the example AI application folder of this course section, there are reference examples that can be directly imported and used.
• In the Dify Studio on the homepage, click "Import DSL File".

• Select the example AI application Complex agent.yml in the folder of this course section, then click "Create".

• You can see the workflow content as shown below.

• In the workflow, Question Classifier is a module driven by an AI large model. Its function is to classify user questions into math questions, physics questions, and other questions. If the user's input is suitable for math or physics related questions, different branch AI large models will be called to answer. The classification content of Question Classifier is as follows:

• The LLM-1 branch is used to answer math-related questions. Its prompt and settings are as follows:

• The LLM-2 branch is used to answer physics-related questions. Its prompt and settings are as follows:

4. Visual Workflow Debugging

• If you need to debug and test the effect of the workflow, click "Preview" in the upper-right corner, then enter a question in the pop-up dialog to test.

• At the same time, the workflow will display the branch the data flows through in real time, making it convenient to debug the workflow.

• If you need to extend the workflow, click the + sign on the left, where there are multiple Dify preset tools and modules.

5. Accessing the AI Agent Application

• After completing the AI application orchestration, click "Publish Application" to save the configuration. Then click "Copy URL" or "API access credentials" to access the created AI application through the web page or backend service API.