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

# Data Augmentation

> Automatically expand your datasets from a handful of examples into comprehensive, realistic datasets

> [!WARNING]
>
> You are reading the docs of Galtea 5.16. [Read this page in the latest version](https://docs.galtea.ai/concepts/product/dataset/data-augmentation.md).

## What is Data Augmentation?

Data Augmentation is a Galtea feature that **automatically generates new test cases from a small set of examples**. Instead of manually writing hundreds of test cases, you provide a few representative samples and Galtea produces a diverse dataset that preserves the realism, structure, and domain-specific patterns of your original data.

It is especially useful when you already have a baseline set of test cases and want to expand your coverage without the overhead of manual creation.

> [!NOTE]
>
> Data Augmentation works with all dataset types — [Accuracy](https://docs.galtea.ai/v5.16/concepts/product/dataset/accuracy-datasets.md), [Security & Safety](https://docs.galtea.ai/v5.16/concepts/product/dataset/security-datasets.md), and [Behavior](https://docs.galtea.ai/v5.16/concepts/product/dataset/behavior-datasets.md). When triggered, your [dataset](https://docs.galtea.ai/v5.16/concepts/product/dataset.md) enters the `AUGMENTING` status while new cases are being generated.
>
> One limit: a dataset whose test cases carry audio or file inputs cannot be augmented. The engine cannot create a new attachment, so every generated test case would point at the same one.

<video controls muted playsInline src="https://mintcdn.com/galtea/CrMHwvQ68HoN2-Q1/videos/data-augmentation-demo.mp4?fit=max&auto=format&n=CrMHwvQ68HoN2-Q1&q=85&s=5290918cf427513935e0813144e78e67" />

## How It Works

Galtea goes beyond simple paraphrasing or template filling. The augmentation engine analyzes your existing test cases to understand their structure, detect domain-specific patterns, and plan diverse scenarios — then generates new cases that are realistic, structurally consistent, and cover different ground from each other.

The result is a larger, high-quality dataset in minutes instead of days, with full control over the output size.

The engine reads a [field rule](https://docs.galtea.ai/v5.16/concepts/metric/json-field-match.md#field-rules) in an expected output as plain JSON text.

## How to Use Data Augmentation

You can augment any existing dataset without audio or file inputs directly from the [Galtea dashboard](https://platform.galtea.ai/):

**1. Navigate to your dataset**

Pick **Datasets** in the sidebar and select the dataset you want to augment. The sidebar keeps **Datasets** behind **More** unless you turn on **Show all tabs** in your settings.

**2. Click Augment Test Cases**

Use the **Augment Test Cases** button to start the augmentation process.

**3. Wait for generation**

Your dataset will enter the `AUGMENTING` status. Once complete, the new test cases will be added to your dataset.

## Related

> [card] [Datasets](https://docs.galtea.ai/v5.16/concepts/product/dataset.md)
>
> Learn about datasets and dataset types in Galtea.

> [card] [Test Cases](https://docs.galtea.ai/v5.16/concepts/product/dataset/test-case.md)
>
> Understand the structure of individual test cases.


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.