Currently, the Guru connector uses board and broad groups APIs, which are deprecated concepts in Guru, and have no UI associated with them in the current Guru web application. The groups feature has been replaced with the more logical and flexible folders concept, where a collection has folders and cards, and a folder has folders and cards.

Guru collections tend to be organized into folders for logical partioning/grouping of cards (at our company). 

Fetching Guru boards via board API calls, today, leads to stale bits of card relationship data from Guru which do not represent the current card relationships in our Guru content.

With the current connector, there is no way to infer folders from the data in the resultant database. That means that the best a user can do with this data is pull all cards associated with a collection, with no inter-card organizational relationships present.

Specificially, in order to create useful RAG vector stores in databricks, I need to pull from specific folders within Guru collections. This is currently impossible with this connector, which is unfortunately putting me in a position to mandate for an alternate data aggregation solution, not Fivetran, for AI at Hinge Health.