Learning instances overview

A learning instance is a configuration where users can define the data to extract (form and table fields), how to extract the data (selecting an appropriate model, defining keywords, queries, regex, and so on), and how to validate the extracted data (data types and rules).

When you create a learning instance, it is saved in your private repository so that you can upload and train sample documents. After the learning instance meets your requirements, make it available for other users to use.

See Publish the learning instance to production.

Private and public modes

Learning instances lifecycle includes two primary phases: training and production. The training phase operates in private mode where users configure and refine the learning instances. The production phase operates in public mode where the trained learning instances are deployed for operational use.

Private mode
This mode is used to train the learning instances by creating, editing, and testing sample documents to automate document processing. After the training is complete, users deploy the learning instances for other users to use in production.
Public mode
In this mode, the learning instances are made available for production use. Users with appropriate permissions can use the deployed learning instances to upload new documents for automatic document processing. Documents that require validation are placed in the validation queue, where multiple users can validate them simultaneously. Also, users can manage and assign approval tasks to other users.

Components

When you create a learning instance, the system creates a default end-to-end workflow that consists of several components. Each of these components has specific tasks associated with the workflow. For example, the extraction bot is used for data extraction. Depending on your use case or business requirements, you can modify the default configuration of these components. For example, if you want to connect to a system of record for validation or for record updates, adding approval steps or notifying users about new tasks, you can edit the Automation Co-Pilot process to include these actions.

You can view these components by navigating to Automation > Document Workspace Processes > li-name folder. See Document Automation processing workflow.

Automation Co-Pilot process
A default process is created for a learning instance to implement the document processing workflow. Depending on your requirements, you can add or remove steps from this process.
Extraction bot
This bot is created to extract data from a document uploaded to Document Automation and to configure BYOL for third-party extraction providers such as generative AI and Google Document AI.
Download bot
This bot is created to download data from server to the output folder.
Automation Co-Pilot form
This form is created to define input fields that will be used to initiate process requests.

Users can view the list of learning instances available for them to use on the Learning Instances page by navigating to Manage > Learning Instances.

Next steps

See Create a learning instance in Document Automation.