Skip to main content
AI transformation components let you use large language models (LLMs) and other AI functions on the data in your cloud data warehouse as part of a transformation pipeline. Each component uses the AI functions that your warehouse provides—such as Snowflake Cortex, Databricks AI functions, or Google BigQuery AI functions—so you can analyze and enrich rows of data without leaving Designer. The components available depend on your cloud data warehouse. To call LLMs and other AI services from an orchestration pipeline, read AI orchestration components.

Use cases

AI transformation components are useful when you want to apply generative AI to the rows of a table, such as to enrich, classify, or search your data, and then pass the results to the rest of your pipeline. Use AI transformation components to:
  • Generate text from a prompt, for example to summarize, translate, or correct the values in a column.
  • Classify rows into categories, or analyze the sentiment of text.
  • Extract specific information from unstructured text, or from documents such as PDFs and images.
  • Mask sensitive entities in text.
  • Compare two pieces of text to calculate how similar they are in meaning.
  • Check each row against a yes/no condition, or rate each row against scoring criteria that you write.
  • Convert text into vector embeddings, and search those embeddings to find the most relevant data.