Create an Agent with the Agent Builder

The AI Agent Builder simplifies the process of creating AI Agents, making it accessible whether you're a data expert or just getting started. Follow this guide to create and configure your first Agent effectively. #

Navigating to the AI Agent Builder

  1. Access the Agents Library

    • From the Main Navigation bar, under "Marketing Apps," select "Agents."
    • This brings you to the Agents Library, where you can view existing AI Agents created by your team or shared by NinjaCat.
    • Use the search function to find specific Agents, mark favorites, and check associated Dataset providers.
  2. Create a New Agent

    • Click "New Agent" in the top right corner to start building a new Agent.

Define Your Goal

Before configuring your Agent, clarify the task you need it to perform. Examples include:

  • Identifying poor-performing keywords.
  • Monitoring ad spend.
  • Generating report summaries.
  • Comparing datasets for discrepancies.

Having a clear objective will streamline the setup process.

Start Building

  1. Interact with the Agent Builder

    • Start a conversation by explaining your requirements. Example prompts:
      • "Analyze ad cost data from the last 30 days and recommend budget adjustments."
      • "Identify Google Ads campaigns that have stopped serving impressions."
      • "Research competitors and analyze SEO keyword opportunities."
      • "Compare two CSV files to find campaign name differences."
    • Click Send to let the Agent Builder configure your Agent automatically.
    • The configuration includes prompt creation, profile picture generation, naming, and description setup.
  2. Modify the Agent’s Instructions

    • Use the "General" tab to view and manually adjust the prompt.
    • Continue refining by messaging the Agent Builder.
  3. Enable Additional Tools (if needed)

    • Some tools are pre-enabled and locked for foundational functions.
    • Optional tools:
      • Code Interpreter – Runs code, analyzes data, performs calculations.
      • Image Generator – Creates images for ads, websites, and marketing.
      • Web Researcher – Finds relevant industry data online.
      • Web Scraper – Extracts website content for analysis.

Set Visibility Preferences

Choose how the Agent will be accessed:

  • Private – Only visible and editable by you (except for NinjaCat employees).
  • Chat Only – Others can interact with it but cannot edit.
  • Public – Fully accessible for editing and use within your agency.

Choose Your Dataset

The AI Agent Builder will recommend relevant Datasets based on your goal. You can:

  • Select from suggested Datasets.
  • Manually add Datasets via the "Knowledge" tab.

Manually Adding a Dataset

  1. Select a Dataset and View (if applicable).
  2. Name the Dataset – Assign a friendly reference name.
  3. Provide a Description (optional).
  4. Review Available Columns – Preview dataset structure.

NinjaNote:

  • Use the "Preview Dataset" icon to inspect data in the Data Explorer.
  • Assign only necessary data to optimize Agent performance.

Test, Tweak, Repeat

  1. Run a test using the Agent Preview.
  2. Evaluate the output:
    • Does it align with your goal?
    • If needed, adjust the prompt for clarity.

Save and Deploy

  • Once satisfied, save your Agent.
  • Start integrating it into your workflow for automated insights and analysis.

Additional Tips for Success

  • Start Simple – Begin with basic tasks and refine over time.
  • Use Examples – Provide sample data to improve accuracy.
  • Experiment – Adjust prompts and settings to fine-tune performance.
  • View and Edit Prompts – Access the "General" tab to see or modify the AI-generated instructions.

By following these steps, you’ll have a functional AI Agent tailored to your needs. In the next module, we’ll explore advanced techniques for optimizing AI prompts.