> Why not ask DataLake team to create one more Kafka/Kinesis topic for your team?

That plus aggregating and some processing it's pretty much what a FS is.

The DataLake is more like a raw data storage and the FS can provide better (and more contextualized ) info on top of that.

As en example: let's say we have all the 5 orders from a given customer in DL. In FS, we can have an attribute/feature that filters and counts the orders done during lunch (calculating and filtering the orders based on their created_at timestamp). That way, we have FS as a source of truth and it's clear to the company what are the rules that define the total_orders_lunch feature.