Skip to content

Create Your First Pipeline

Introduction

ELITEA Pipelines are powerful workflow automation tools that visually design and execute sequences of states and actions. Each pipeline is designed to handle specific tasks or sequences of tasks by connecting various nodes such as conditions, decisions, loops, and tool integrations. This guide will walk you through creating your first pipeline from start to execution.

Step 1: Navigate to the Pipelines Menu

  1. Open the Sidebar:

    • From the main platform navigation, locate and click Pipelines in the sidebar.

  2. Access the Pipelines Dashboard:

    • You'll see a list of all existing pipelines for your project (if any).
    • If this is your first time, the list will be empty.

Step 2: Create a New Pipeline

  1. Start Pipeline Creation:

    • Click the + Create button located at the top of the main sidebar.

  2. Configure General Information:

    • Name: Enter a unique, descriptive name for your pipeline (e.g., "Data Processing Workflow", "Report Generation Pipeline").
    • Description: Provide a clear description of what your pipeline will do.
    • Tags (optional): Add relevant tags by typing tag names or selecting from existing ones.

  3. Save Initial Pipeline:

    • Click Save to create your initial pipeline.
    • This creates the "latest" version of your pipeline.

  4. Configure Optional Features:

    • Welcome Message (optional): Add a message that users see when they first interact with your pipeline.
    • Conversation Starters (optional): Create predefined prompts to help users get started quickly.

  5. Save Your Configuration:

    • Click Save again to preserve your additional configurations.

Pipeline Creation Interface

Step 3: Add Toolkits, Agents, or Pipelines

After saving your initial pipeline, the TOOLKITS section becomes available with three integration options to enhance your pipeline's functionality.

Add a Toolkit

  1. Access Toolkits Section:

    • In your pipeline configuration, scroll to the TOOLKITS section.

  2. Add a Toolkit:

    • Click the + Toolkit button.
    • Select Existing Toolkits: Browse and choose from available toolkits in the platform's toolkit library, or
    • Create New Toolkit: Click the + Create new button to build a custom toolkit for your specific requirements.

Reference

For detailed information on creating and managing toolkits, refer to the Toolkits Menu Guide.

Add an Agent

  1. Click the + Agent Button:
    • This allows you to incorporate existing agents into your pipeline workflow.

Add a Pipeline

  1. Click the + Pipeline Button:
    • This allows you to nest other pipelines within your current pipeline.

Pipeline Toolkits Configuration

Step 4: Design Your Pipeline Workflow

Using the Flow Designer

  1. Access Flow Designer:

    • Navigate to the Configuration tab.
    • Select the Flow tab to visually design your pipeline.

  2. Add Nodes:

    • Click the + icon to add new nodes.
    • Choose from available node types: Agent, Condition, Decision, Function, LLM, Loop, Loop from Tool, Pipeline, Router, State Modifier, Tool, and Custom.

  3. Connect Nodes:

    • Drag and drop connections between nodes to establish workflow logic.
    • Use the End node to define pipeline completion.

  4. Customize View:

    • Zoom in or out for better navigation.
    • Adjust the view for managing complex workflows.

Pipeline Flow Designer

Using the YAML Editor (Optional)

  1. Switch to YAML:

    • Click the YAML tab for advanced code-based configuration.

  2. Define Workflow:

    • Write complex workflows, conditions, and logic using YAML syntax.
    • Fine-tune node configurations and set advanced parameters.

  3. Validate Syntax:

    • Ensure your YAML syntax is correct before saving.

Step 5: Test and Execute Your Pipeline

  1. Navigate to Run Tab:

    • Click on the Run tab to access the execution environment.

  2. Select AI Model:

    • Choose an appropriate AI model from the dropdown (e.g., gpt-4o-2024-11-20).

  3. Adjust Parameters (optional):

    • Temperature: Control creativity level (0.1 for consistent, 1.0 for creative).
    • Top P: Adjust word selection diversity.
    • Max Completion Tokens: Set response length limit.

  4. Start Execution:

    • Use a Conversation Starter (if configured), or
    • Type your question or command directly.
    • Use simple commands like "Go", "Start Generating", "Execute", or "Run it".
    • Click Send to execute your pipeline.

Pipeline Execution Interface

Monitor Execution:

  • View the pipeline execution flow and results in real-time.
  • Track execution progress and analyze outputs.

Post-Execution Actions:

  • Copy Output: Click the copy icon to save results.
  • Regenerate: Click regenerate if output isn't satisfactory.
  • Continue Dialogue: Type follow-up questions or commands.

Version Management:

  • Save: Update the current "latest" version or the current named version
  • Save As Version: Create a new named version
  • Publish: Submit for approval to make publicly available

Step 6: Add Your Pipeline to Conversations

The primary way to use your pipeline is by adding it to conversations for collaborative work.

  1. Navigate to Chat Menu:

    • Go to the Chat section from the main sidebar.

  2. Start a New Conversation:

    • Click + Create to start a new conversation.
    • Or select an existing conversation where you want to add your pipeline.

  3. Add Your Pipeline:

    • At the bottom of the chat, you can see the switch to assistant icon.
    • To switch assistants (e.g., select a Pipeline):
      • Click the Switch assistant icon.
      • In the "Frequently Used" list, click your pipeline name (e.g., Data Processing Workflow) to select it.
      • Alternatively, type # followed by your pipeline name (e.g., #Data Processing Workflow) in the input box to quickly select an assistant
    • Alternative Method: Click the + button on the PARTICIPANTS panel next to the Pipeline section to add your pipeline directly to the conversation.

  4. Interact with Your Pipeline:

    • Type your question or request and send it.
    • Your pipeline will respond based on its workflow design and available toolkits.

Conversation Benefits

Using pipelines in conversations allows for collaborative workflows where human team members and AI pipelines work together, maintaining context and continuity throughout the discussion.

Add to Conversation

Reference

For detailed instructions on creating a conversation, refer to the Create first Conversation guide.

Next Steps

Now that you've created your first pipeline, consider:

  • Adding more toolkits to expand integration capabilities
  • Creating complex workflows with multiple node types
  • Building nested pipelines for modular workflow design
  • Setting up monitoring for performance optimization
  • Publishing versions for team collaboration

Best Practices

  • Plan your workflow before adding nodes for clear structure
  • Use descriptive names for nodes and pipelines
  • Test iteratively during development
  • Validate data flow between nodes with sample data

Related Documentation

For more detailed information, refer to: