One thing that could probably improve is giving users more transparency into confidence scores and extraction reasoning. For sensitive workflows like legal or financial processing, visibility into why certain outputs were generated can help teams trust the system more
Parsewise
Thank you for the review @rohit_cpp07 !
You are most welcome to set up a free account to see more details on the extraction reasoning and visibility of results:)
Here is an example from our sandbox demo environment.
You can see the reasoning for the final single value from different sources.
For each of the underlying values there is both reasoning and tracing show.
Here are some screenshots for reference: