The CSV Loop feature enables your agent to process data systematically by iterating through each row or column of a CSV file. Each iteration triggers a separate agent execution, making it ideal for batch processing operations.

This powerful capability is especially valuable for:

  • Data cleaning and standardization
  • Large-scale content generation
  • Data enrichment and validation
  • Automated analysis across multiple records

​
Set Up CSV Looping

Follow these steps to configure and use the CSV Loop feature:

  1. Navigate to Behavior Settings: In Agent Studio, go to Behavior.

  2. Enable CSV Loop:
    Toggle on the Loop over CSV option.

    Enabling this feature activates the CSV upload functionality in the Playground.

  3. Upload Your CSV: In the Playground, you will now see a CSV upload option. Upload your desired CSV file.

  4. Configure Iteration: Select whether you want the agent to loop over rows or columns.

  5. Start the Loop: Configure your agent’s message or prompt as needed. Send the message in the chat to initiate the loop operation.

​
Monitor Loop Progress

You can track the progress of your CSV loop as it runs:

  • Real-time Tracking: Monitor loop progress directly in the Playground chat interface while it’s active.

  • Session Limitation:

    If you close the Playground and return later, the real-time loop progress will not be visible in the chat history.

  • Completion Notification: Once the loop completes, you’ll receive a summary notification in the Notification Center.

​
Common Use Cases

The CSV Loop is powerful for various tasks, including:

​
Data Cleaning and Standardization

Scenario: Process a CSV file containing inconsistent or incomplete data.

Input Data Issues Examples:

  • Inconsistent address formats
  • Missing company websites
  • Varied job title formats
  • Incomplete contact information

Agent Actions:

  • Address Processing: Parse and reformat addresses into a standard structure (e.g., Street, City, State, Zip).
  • Phone Number Standardization: Convert phone numbers to a consistent format.
  • Website Enrichment: If CompanyName exists but Website is missing, search the web for the official company website.
  • Industry Classification: Infer Industry or CompanySize categories based on JobTitle and CompanyName.

Output: A cleaned and enriched dataset with standardized information.

​
Product Description Generation

Scenario: Create unique product descriptions from attribute data.

Input Data Examples:

  • ProductName
  • Features (list or descriptive text)
  • Material specifications
  • TargetAudience demographics
  • ProductCategory

Agent Actions:

  • Feature Analysis: Analyze product features and benefits.
  • Audience Targeting: Tailor language and messaging to the specified target audience.
  • Description Generation: Create compelling product descriptions that highlight benefits derived from features.
  • SEO Optimization: Generate meta descriptions and relevant keywords.
  • Format Optimization: Structure content with bullet points for key features and benefits.

Output: Professional product descriptions ready for e-commerce platforms or marketing materials.

​
Best Practices

  • Data Preparation: Ensure your CSV file is properly formatted with clear column headers.
  • Agent Configuration: Test your agent’s behavior on a small sample of data before processing large datasets.
  • Progress Monitoring: Keep the Playground open during execution to monitor real-time progress.
  • Result Validation: Review the completion summary and validate the outputs as needed.