Transforms¶

Data preprocessing and transformation are essential for designing effective AI agents. To prepare raw or semi-structured data for interactions, it often needs to undergo a series of transformations.

The transforms functionality allows you to adjust and refine AI reasoning and output for the AI agent. By using transforms, you can tailor how your AI agent processes input data and generates responses, aligning them with your specific needs and requirements.

In Alan AI, transforms enhance AI performance in the following ways:

Improving AI reasoning: you can adjust how the AI agent processes and interprets input data to refine its decision-making logic.
Enhancing AI output: you can modify and format AI responses to make them clearer and more aligned with user expectations.

Alan AI supports the following formats for data transformations:

Input format: plain text, markdown-formatted text, HTML, JSON and JavaScript
Output format: plain text, markdown-formatted text, HTML, JSON and JavaScript

To use transforms in Alan AI, you need to:

Create a transform with a set of examples that will instruct Alan AI on how to adjust and format input data. For details, see Transform examples.
Apply the created transform.

To get started with transforms, read the following sections:

Transform configuration

Understand what parameters you can use to adjust the AI model behavior and output

Transform instructions and examples

Find out how to define instructions and examples in transforms

Static corpus transforms

Learn how to adjust AI reasoning and output for static data corpuses.

Dynamic corpus transforms

Learn how to adjust AI reasoning and output for dynamic data corpuses.

Puppeteer transforms

Learn how to adjust AI reasoning and output for corpuses created with the Puppeteer crawler.

Output transforms