​
Key Characteristics of Dynamic Columns

  • Automated Value Generation: Values are computed dynamically rather than manually entered.
  • Regenerable Data: Dynamic columns can refresh their values when the underlying logic or data changes.

​
How Dynamic Columns Work

  1. User selects a dynamic column type and configures its settings.
  2. The system uses parallel execution to compute values efficiently for large datasets.
  3. Each row in the column receives a computed value based on the logic defined.

​
Why Use Dynamic Columns?

  • Efficiency: Reduces manual data entry and updates values automatically.
  • Scalability: Works efficiently on large datasets with thousands of rows.
  • Flexibility: Supports various logic types, from simple conditions to external data integrations.
  • Data Consistency: Ensures uniform value generation across the dataset.

​
Supported Methods:

  • Run Prompt: Run a LLM prompt which can also utilize data from static columns to get the desired output
  • Vector Retrieval: Connect to a Vector Database, and retrieve the top-k chunks for a particular query.
  • Entity Extraction: Automatically extract named entities like people, organizations, or locations from the static columns using a specified model.
  • JSON Key Extraction: Parse a JSON field to extract specific keys or values.
  • Custom Code Execution: Write and execute Python code for transformations or complex operations.
  • Text Classification: Assign categories or labels using a specified models.
  • API Calls: Generate a column and new entries for every row by making API calls to a specified endpoint.
  • Conditional Logic: Apply different actions to your data based on specified conditions, allowing for branching logic.
By leveraging dynamic columns, users can automate data transformation, fetch external insights, and apply complex logic, making their datasets more powerful and adaptive