Issue saving manual corrections to machine learning

I followed the video and explanation from the new AI Invoice post. I was able to get a demo working on invoices of our own from following the post where I can open each invoice in the validation station.

The video and steps list skim over some of the parts needed to make your own workflow, but I still cannot find enough information in the documentation in order to make sense of how to calibrate the extractor to save the manual selections in order to process the same invoice template correctly the next time it processes. Why after manually selecting the fields in the validation station, does the next iteration with the same invoices not find the correct fields? Is this something that only happens over time, or how does it work?

Are there any resources available with a bit of a “deeper dive” into the implementation and details behind this sweet new feature?

Hi @mtu

Try this

Machine Learning with UIpath RPA
Thanks
Ashwin.S