I'm trying out the beta version of Box AI for Hubs.
I expected this feature to be able to combine information from multiple documents within Hubs to generate an answer.
However, at the moment, even though there is relevant information in the documentation within Hubs, only general answers are returned, and no satisfactory answers have been obtained.
As it is, it would be difficult to deploy it internally, so we would like to see further improvements in the accuracy of AI.
Please note that all content is written in Japanese.