Get to Know the Data Integration Agent

Learning Objectives

After completing this unit, you’ll be able to:

• Describe the role of CLAIRE GPT in Cloud Data Integration.
  
• Identify the prerequisites needed to create and run extract, load, transform pipelines.
  
• Explain the two methods to create extract, load, transform pipelines in CLAIRE GPT.
  

Before You Start

Before you start this badge, consider completing this recommended content.

• CLAIRE GPT Basics
  
• CLAIRE GPT Interface and Navigation
  
• CLAIRE GPT Discovery Agent
  

Turn Analysis into Pipeline

Data analysts often struggle to efficiently transition from answering one-off questions to creating the repeatable data pipelines that other systems require.

So far, Ana at Cloud Kicks has used CLAIRE GPT to discover trustworthy datasets and analyze them directly in conversation. But sometimes an answer isn't the end goal—it's the start of a repeatable data flow that other systems and reports rely on. That's where the Data Integration Agent comes in.

In this unit, you follow Ana as she moves from asking questions about her data to building a pipeline. The extract, load, transform (ELT) pipeline turns her analysis into a reusable data asset in Cloud Data Integration (CDI).

The Role of the Data Integration Agent

The Data Integration Agent helps you create ELT pipelines that transform data, entirely through conversation. Instead of designing a mapping from scratch in Data Integration, you describe what you want in natural language. The agent builds the pipeline logic for you. You can open, modify, and run that pipeline as a mapping in CDI.

The agent works autonomously. It interprets your request, decides how to structure the transformation, such as aggregating, filtering, or joining data, and generates the resulting mapping. You can access this capability through CLAIRE GPT or CLAIRE Copilot.

Example prompt: “Create a mapping to load customers who have a high frequency of engagement.”

The Data Integration Agent can't create, move, store, or delete existing data. Ana learned this when she asked, “Can you store the results in a new target table?” CLAIRE GPT couldn't complete that request. The agent builds the transformation logic, but you still run, save, and operationalize it in CDI.

Prerequisites for ELT Pipelines

A few things need to be in place before Cloud Kicks analysts can build pipelines with CLAIRE GPT. This setup is similar to what Ana's administrator handled for CLAIRE GPT overall. Here are the requirements.

• Informatica Cloud Data Governance and Catalog (CDGC): The CDGC service must be enabled for the runtime environment your organization uses.
  
• CDI: The CDI service must be enabled for that same runtime environment, so it can actually run the mappings the agent creates.
  
• Extracted metadata: The metadata must already be extracted from the relevant source systems into the catalog, so the agent can only transform data it can already discover.
  
• The Designer role: This role lets you open, save, and run a mapping in CDI.
  

Two Ways to Create a Mapping

CLAIRE GPT can create a mapping in one of two ways, and knowing which one fits your situation saves time.

1. A data exploration prompt followed by a data transformation prompt: Ana first asks a question to explore and confirm the data she wants, such as customers with high engagement. She reviews the results, and then asks CLAIRE GPT to turn that result into a mapping.
   
2. A single data transformation prompt: When Ana already knows exactly what she wants, she can skip the exploration step and ask for the mapping directly. For example, she can request the total number of customers with below-average engagement per month.
   

The single-step approach is faster when you’re confident in a request. The two-step approach lets you confirm the data is correct before you commit to a mapping. Ana is confident, so she chooses a single-step prompt.

Example prompt: “Create a mapping that tells me the total number of customers with a below-average frequency of engagement per month.”

Keep one limitation in mind. A data exploration prompt that relies on a complex SQL query doesn’t always convert cleanly into a mapping. When the conversion doesn’t work as expected, Ana simplifies her question.

Wrap Up

You now know what the Data Integration Agent does, what you need before you use it, and how to start mapping. In the next unit, you build a pipeline from a prompt.