> ## Documentation Index
> Fetch the complete documentation index at: https://futureagi.mintlify.site/llms.txt
> Use this file to discover all available pages before exploring further.

# Create Dynamic Column by Executing Code

> The **Execute Custom Code** feature allows users to create a dynamic column by writing and running Python code on dataset rows. This enables custom transformations, calculations, or data processing based on existing column values.

The **Execute Custom Code** feature allows users to create a dynamic column by writing and running Python code on dataset rows. This enables custom transformations, calculations, or data processing based on existing column values.

By defining a function, users can manipulate row-level data and store the results in a new column.

***

## **1. Select a Dataset**

Before executing custom code, ensure you have selected a dataset from your workspace. If no dataset is available, follow the steps to **Add Dataset** on the Future AGI platform.

***

## **2. Accessing the Custom Code Execution Interface**

To configure a custom column, navigate to your dataset and click the **+ Add Columns** button in the top-right menu. Scroll down to the **Dynamic Columns** section and select **Execute Custom Code** to open the setup panel.

***

## **3. Configuring Custom Code Execution**

* **Name**: Assign a name to the new column where the computed results will be stored.
* **Python Code**: Write a Python function to process row data. The function should be named `main` and accept keyword arguments (`kwargs`) to access column values.
* **Concurrency**: Define how many rows should be processed simultaneously for efficiency.

***

After writing the function, click **Test** to preview the computed values. If the output is correct, click **Create New Column** to apply the function to all rows in the dataset. The newly created column will update dynamically with computed values.

***

## **Best Practices for Custom Code Execution**

* **Use simple, efficient Python logic** to avoid performance issues.
* **Ensure column names are correctly referenced** in the function.
* **Test the function before applying it** to catch errors early.
* **Optimize concurrency settings** for large datasets to balance speed and processing power.

***