Help with Invoice AI setting up separate workflows for extraction

Help Document Understanding
, ,
1

So I have a full workflow working to process invoices using the new Invoice AI Machine Learning activities, and it is working quite well for what we are trying to use it for.

Just one issue, the digitize document and data extraction parts take too long. The user would have to wait several long seconds in between validating each invoice which is really bothersome while actually using/testing this. I got help from someone at UiPath to figure out other parts of this and she suggested that we use Process Orchestration for data extraction and validation.

I tried to follow the documentation linked above but I guess it is over my head.

How could I implement that for a workflow that a person would initiate and does anyone have any examples?

2

Hi there,

You might want to look into Long Running Workflows - with Action Center integration of the validation station - or at least break up your project into

  • one unattended that does the automatic processing (digitization and data extraction included)
  • one attended that only does human validation
  • one unattended that picks up human validation and does post-processing on it.

Coordination between the processes, you can do using orchestrator queues.

I do recommend looking into the Action Center VS integration though :slight_smile:

Check out the latest sample in here: How to use the IntelligentOCR Package

1 Like
3

Ioana,
This is what we ended up doing. Creating a dispatcher to digitize and extract and a performer to open each invoice in the validation station on the user’s machines. Works quite well!

I remember you giving that option when we first started looking into this, thanks!

4

This topic was automatically closed 3 days after the last reply. New replies are no longer allowed.