---
title: 'Creating Chatbots'
description: 'Step-by-step guide to building chatbots in PySpur'
---

# Creating Chatbots in PySpur

This guide will walk you through the process of creating, configuring, and deploying a chatbot in PySpur. We'll build a simple customer support chatbot that can answer questions about a product.

## Creating a New Chatbot

<Steps>
  <Step title="Open the Dashboard">
    Start by navigating to the PySpur dashboard. This is your central hub for managing all your Spurs.
  </Step>

  <Step title="Create a New Spur">
    Click the **New Spur** button in the top right corner of the dashboard.

    ![New Spur Button](/images/chatbots/new-spur-button.png)
  </Step>

  <Step title="Select Chatbot Type">
    In the modal that appears, select **Chatbot** as the spur type.

    <Note>
      PySpur automatically sets up the required input and output nodes with the correct fields for a chatbot.
    </Note>
  </Step>

  <Step title="Name Your Chatbot">
    Give your chatbot a descriptive name, such as "Product Support Bot" or "Customer Help Assistant".
  </Step>
</Steps>

## Understanding the Default Chatbot Structure

When you create a new chatbot, PySpur automatically creates two nodes:

1. **Input Node**: Contains the required fields:
   - `user_message`: The message from the user
   - `session_id`: A unique identifier for the conversation
   - `message_history`: An array of previous messages (automatically populated)

2. **Output Node**: Contains the required field:
   - `assistant_message`: The response from your chatbot

## Adding Intelligence to Your Chatbot

<Steps>
  <Step title="Add an LLM Node">
    Drag an LLM (Large Language Model) node from the sidebar onto the canvas. This will be the "brain" of your chatbot.

    Connect the Input node to your LLM node, and the LLM node to the Output node.
  </Step>

  <Step title="Configure the LLM Node">
    Click on the LLM node to open its configuration panel. Here you can:

    - Select the model provider (OpenAI, Anthropic, etc.)
    - Choose the specific model (GPT-4, Claude, etc.)
    - Customize the prompt template to guide the model's responses

    <Tip>
      Structure your prompt to mention that this is a customer support bot for your specific product or service. Include instructions about tone, personality, and knowledge limits.
    </Tip>
  </Step>

  <Step title="Format Messages">
    In the LLM node configuration, make sure your prompt references the incoming data correctly:

    ```
    You are a helpful customer support agent for [Your Product].

    Previous conversation:
    {{message_history}}

    User: {{user_message}}

    Assistant:
    ```

    This template ensures the LLM sees the conversation history and the latest user message.
  </Step>
</Steps>

## Enhancing Your Chatbot with Additional Nodes

You can make your chatbot more powerful by adding other nodes:

- **Vector Database Node**: Connect to a vector database to retrieve product information
- **Tool Node**: Enable the chatbot to perform actions like looking up orders or tracking shipments
- **Branching Node**: Direct the conversation flow based on user intent
- **API Node**: Connect to external services to fetch real-time data

## Testing Your Chatbot

<Steps>
  <Step title="Open the Test Panel">
    Click the "Test" tab in the right sidebar to open the testing interface.
  </Step>

  <Step title="Set Test Values">
    Enter test values for the required input fields:
    - `user_message`: "What features does your product have?"
    - `session_id`: "test-session-123"

    You don't need to provide `message_history` as it's automatically managed.
  </Step>

  <Step title="Run the Test">
    Click the "Run" button to execute the workflow and see the chatbot's response.
  </Step>

  <Step title="Continue the Conversation">
    To test multi-turn conversation, enter a follow-up message like "How much does it cost?" using the same session ID.

    Notice that the chatbot now has access to the previous messages and can maintain context.
  </Step>
</Steps>

## Deploying Your Chatbot

Once you're satisfied with your chatbot:

<Steps>
  <Step title="Save Your Work">
    Click the Save button to save your chatbot configuration.
  </Step>

  <Step title="Deploy as API">
    Click the "Deploy" button in the top bar. This creates an API endpoint for your chatbot.
  </Step>

  <Step title="Get Integration Details">
    You'll receive API details that you can use to integrate the chatbot:

    - API URL: The endpoint to call
    - Request format: How to structure calls to your chatbot
    - Authentication details: How to securely access your chatbot
  </Step>

  <Step title="Integrate with Your Application">
    Using the API details, you can now integrate the chatbot with:

    - Your website using JavaScript
    - Mobile apps
    - Custom UIs
    - Other backend systems
  </Step>
</Steps>

## Example: Simple Customer Support Chatbot

Here's a simple example of a customer support chatbot for a fictional product:

1. **Input Node**: Receives user questions
2. **Vector Database Node**: Searches product documentation for relevant information
3. **LLM Node**: Formulates a helpful response using the retrieved information
4. **Output Node**: Returns the response to the user

This chatbot can answer questions about product features, troubleshooting steps, and basic account information.

## Next Steps

- **Customize Node Configurations**: Fine-tune your chatbot's behavior
- **Add Authentication**: Secure your chatbot API
- **Implement Analytics**: Track and analyze chatbot usage
- **Add Specialized Nodes**: Enhance capabilities with custom functionality

By following this guide, you've created a basic but functional chatbot in PySpur. As you become more familiar with the platform, you can create increasingly sophisticated chatbots for various use cases.