I have tried using Frontline Copilot, and I must say it is incredibly powerful. It accurately retrieves the necessary information from Japanese-language tables. However, when working with documents (files), I noticed a difference in performance depending on the language of the content. Files written in English were processed effectively, and the responses were accurate. However, when the content was in Japanese, I was unable to obtain any answers.
I assume that file processing has not yet been adapted for multiple languages. Do you have any plans to support Japanese and other languages in the future?
Your observation here is correct - file processing uses AWS textract which is somewhat limited with languages it supports: Best Practices - Amazon Textract “Currently, Amazon Textract supports English, Spanish, German, Italian, French, and Portuguese.”
I am inquiring about any work here to be able to support Japanese in Frontline Copilot file processing with the Tulip Product Team and they will get back to you!
Just an update, we are turning this into a product suggestion so we have it in our backlog. The Product Team will take a look, but there is already other ongoing work for Copilot (including PDF → App!) so improvements to add more languages here will be further out in the timeline.
