The May 2025 release of Data Integration includes the following new features and enhancements.
CLAIRE Copilot
CLAIRE Copilot is available to help you create and summarize mappings in Data Integration and to help you answer questions about Informatica Intelligent Cloud Services:
Create mappings
CLAIRE Copilot can create mappings based on a natural language prompt, making it easier and faster than creating mappings from scratch. You can describe the mapping that you want to create, and CLAIRE Copilot creates it and populates all transformations and transformation properties.
Summarize mappings
CLAIRE Copilot can help you summarize mappings that have been created in your organization. You can open a mapping and ask CLAIRE Copilot to summarize it without manually analyzing each transformation to understand how it's configured.
Ask for help with Informatica Intelligent Cloud Services
To use CLAIRE Copilot, go to the Home page in Data Integration and click the CLAIRE icon to open the conversation window. The following image shows the conversation window:
You can select one of the sample prompts to get started, or you can enter your own prompt. For a list of examples and more information, see CLAIRE Copilot for Data Integration.
Hierarchical mappers
In a hierarchical mapper, you can use an intelligent structure model as the target schema with the following output formats: relational, JSON, XML, or JSON Lines.
For more information, see Components.
Intelligent structure models
When you generate output groups using entity relationships, you can now use the maxOccurs schema attribute to create multiple individual output fields.
For more information, see Components.
Mapplets
When you edit a mapplet in the Mapplet Designer, you can synchronize the changes to the affected assets that use the mapplet.
For more information, see Components.
Source control
You can compare two versions of a source-controlled asset from the Explore page. You can compare the current version to an older version or compare two older versions.
For more information, see Asset Management.
Transformations
This release includes the following enhancements for transformations:
Chunking transformation
The Chunking transformation includes the following enhancements:
- You can process text to make the data cleaner and semantically more consistent for vector embedding using the following text processing functions:
- Cleanse text
- Remove diacritics
- Check spelling
- Lemmatize
- Remove stop words
- The Chunking transformation can run in a serverless runtime environment on Microsoft Azure.
Lookup transformation
You can configure cached Lookup transformations that read data from a single source object to include only required fields in the lookup query. When you do this, the lookup query doesn't include fields that are excluded from the downstream transformation in the mapping or mapplet.
Hierarchy Builder transformation
You can write large amounts of hierarchical data to cloud-based targets, such as Microsoft Azure Data Lake Storage Gen2 and Amazon S3 V2, through the Hierarchy Builder transformation using the binary output format.
SQL transformation
You can edit the Precision and Scale metadata fields in SQL transformations for relational connections and ODBC connections that support stored procedure.
Structure Parser transformation
You can use the Structure Parser transformation for COBOL streaming to parse large data sets using COBOL copybooks.
Vector Embedding transformation
The Vector Embedding transformation includes the following enhancements:
- You can connect to your own vector embedding model on Azure OpenAI by selecting or creating a Large Language Model connection.
- The Vector Embedding transformation can run in a serverless runtime environment on Microsoft Azure.
For more information about transformations, see Transformations.
