IDP Automation
The IDP action category introduces Intelligent document processing capabilities using integrated Large Language Models [LLM's] such as Gemini, to extract data. This feature is optimized for processing documents that have unstructured, semi-structured content or complex layouts (e.g., PDFs and images).
The main benefit is the ability to extract text accurately without having to manually configure or indicate the data areas in the file.
Supported RPA agent platforms: Windows
Supported LLM's : Gemini [2.5 Flash]
Supported LLM's : Gemini [2.5 Flash]
To use this action, you must first enable and configure Zia in your settings page, as detailed in the help document here.
Usage limits for this action depend on the token limits of your specific LLM service plan
Usage limits for this action depend on the token limits of your specific LLM service plan
Common use cases:
Process invoices and purchase orders: Extract fields such as invoice date, vendor name and total from highly diverse formats received from different vendors.
Handle customer onboarding documents: Extract data from multiple versions of utility bills, contracts, and government IDs that have different layouts due to region, branding and others.
Analyze legal and compliance documents: Extract key data fields such as termination clause, effective date, party names etc., from unstructured contracts.
Process complex reports: Extract specific information from non-standard reports or medical records (e.g., diagnosis codes, procedure dates, financial figures) from unstructured multi-page medical records or financial reports.
Extract text from file using AI
This action allows you to extract data from PDF or image files without manually configuring each field. You can choose to extract the entire document's content, or limit the extraction using a text prompt.
Action configuration
Variable Name: Specify the name of the output variable that will store the extracted text from the file. This variable can then be used in subsequent steps of your automation flow.
File path to define output variables: The path to the sample file (e.g., an invoice) that the AI analyzes to define the exact output variables that you will need.This is a one-time setup to establish the variables that can be used in subsequent actions.
Choose how to extract text: Select either entire file or based on a text prompt.
Entire file: The AI would process the entire file and generate output variables.
Based on a text prompt: The AI extraction is based on the specific prompt given. (e.g., Extract bill to, invoice date and total from the invoice.)
Extract text using AI: This will process the file given in the file path for output variables using AI, based on the selection you've made, and get the required output variables. You can verify the output values received. You can use these outputs in the subsequent actions of your automation flow.
Execution file path: The path to the file that the bot will process when the RPA flow runs. You can use a static path or a variable (e.g., a file trigger variable) as the input. The processed values will be mapped to the output variables that were generated in the previous step using Extract text using AI button.
Once you have finalized your configuration, click Done.