I am trying to Classify a set of Customer Loan Documents and extract the values from the documents (using Generative classification & Extraction) and pass the values to an Agent for analysis - what should be the way to go about it. How do we pass the data variables for classification as well as Extraction. What data types should we be working with
Dear @rbanik,
The output of the extraction activity is an dataset. UiPath agents dont accept dataset as an input argument.
Hence, first you have retrieve the individual fields from the dataset.
Create input argument in agent and pass the input fields retrieved in first step to the agent.
To invoke agent in RPA process use Run Agent activity. After selecting the agent, pass the input arguments to the agent.
Note: The input argument created in agent builder are not automatically fetched in Studio. Hence, you need to be careful while passing them. Names should matches exactly.
Hope this solves your queries. Dont forget to mark as solution.
Br,
NikZ
Hi NikZ,
- I was thinking of calling the Classification & Extraction RPA Tool from my agent. there are multiple Doc types that need to be classified and extracted (Pay Stubs, W2, Loan Application form, Credit reports, etc.) - which involve a ton of data.
If the extracted data for each Doc Type is passed as Out Argument (collection - for each Doc Type) from the RPA, when RPA is called from Agent, does the agent have intrinsic ability to get and understand the Collection data from Out Arguments of RPA?
Or do I need to call the agent from within the RPA and create separate input arguments for each of the data elements involved?
- Is it possible to create global variables in an Agentic Orchestration process and pass which all the building blocks (RPA, Agents, etc.) within the Agentic Orchestration process can readily access - in this case pass the extracted data from each Doc type (out arguments of RPA) without having to create Arguments in Agent explicitly