跳至主要内容

Building a Basic AI Agent

In this tutorial, we'll walk through the process of creating a basic AI agent using Asgard. This agent will be able to answer questions and perform simple tasks.

Prerequisites​

Before starting this tutorial, make sure you have:

  • An Asgard account
  • A project created in Asgard
  • Basic familiarity with the Asgard interface

Step 1: Create a New Workflow​

  1. Navigate to your project dashboard
  2. Click on "Workflows" in the left sidebar
  3. Click the "Create New Workflow" button
  4. Name your workflow "Basic Agent"
  5. Select "Conversational Agent" as the workflow type
  6. Click "Create"

Step 2: Configure the Agent​

  1. In the workflow editor, locate the "Agent Configuration" panel
  2. Set the following parameters:
    • Agent Name: "Help Assistant"
    • Description: "A helpful assistant that answers questions"
    • Model: Select GPT-4 or another appropriate model
    • Temperature: Set to 0.7 (balanced between creativity and accuracy)
    • System Message: Enter instructions for your agent's behavior

Step 3: Add Input Processing​

  1. Drag an "Input Processor" component from the component library
  2. Connect it to the "Start" node
  3. Configure the input processor to handle user messages

Step 4: Add Response Generation​

  1. Drag a "Response Generator" component from the component library
  2. Connect it to the "Input Processor"
  3. Configure the response settings to generate helpful answers

Step 5: Add Output Formatting​

  1. Drag an "Output Formatter" component from the component library
  2. Connect it to the "Response Generator"
  3. Configure the output format (e.g., text, markdown, etc.)
  4. Connect the "Output Formatter" to the "End" node

Step 6: Test Your Agent​

  1. Click the "Test" button in the top-right corner
  2. Enter a test question in the chat interface
  3. Verify that your agent responds appropriately
  4. Make adjustments to your workflow as needed

Step 7: Deploy Your Agent​

  1. Click the "Deploy" button in the top-right corner
  2. Select your deployment target (e.g., Web, API, etc.)
  3. Configure deployment settings
  4. Click "Deploy" to make your agent available

Next Steps​

Congratulations! You've created a basic AI agent using Asgard. In the next tutorial, we'll explore how to enhance your agent with additional capabilities, such as accessing external data and performing actions.