Skip to main content
Public preview
Production use of this feature is available for specific editions only. Contact our sales team for more information.
The Document AI Predict transformation component extracts data from documents, such as PDFs and images. It invokes the Snowflake document predict function and lets you call Document AI models from the context of a pipeline.

​
Use case

This component can be used to extract data for a range of purposes. For example, it is useful in industries such as finance, logistics, and healthcare, where documents like invoices, contracts, and patient records are often available in different formats, including PDFs and scanned images.

​
Video example

​
Prerequisites

  • Files you wish to process must be in your Snowflake stage.
  • You must have created and configured a Document AI build model in Snowflake already. For more information, read Set up the required objects and privileges.
  • The Document AI Predict component requires that your file’s relative path and presigned URL be available in a Snowflake table. Use this example query to populate your table:
select
    relative_path,
    GET_PRESIGNED_URL(@<stage_name>, relative_path) presigned_url
from directory(@<stage_name>);
The final step in the Snowflake tutorial, Create a document processing pipeline, isn’t required in .

​
Properties

​
Name
string
required
A human-readable name for the component.
​
Database
drop-down
required
The Snowflake source database. The special value [Environment Default] uses the database defined in the environment. Read Databases, Tables and Views - Overview to learn more.
​
Schema
drop-down
required
The Snowflake source schema. The special value [Environment Default] uses the schema defined in the environment. Read Database, Schema, and Share DDL to learn more.
​
Model Build Name
string
required
The build name of the Document AI model. Read Prepare a Document AI model build to learn more.
​
Model Build Version
string
Optionally specify the version of the model to use. If not set, this parameter will default to the latest version.
​
URL Column
drop-down
required
The source column containing the presigned URLs of the staged files the model should act on.Presigned URLs let the user bypass the authentication and sign-in process.As part of the Document AI API, provide a presigned URL to the document you want to run the model against. Document AI then uses the presigned URL to fetch the intended document.
​
Include Input Columns
boolean
required
  • Yes: Outputs both your source URL columns and the prediction columns. This is the default setting.
  • No: Only includes the new prediction columns.