After reading the docs of flyte agent it is still ...
# flyte-support
c
After reading the docs of flyte agent it is still unclear to me what purpose it serves. For example, if I want to call supabase from my task why shall I wrap that call with an agent vs just calling it from the task directly?
f
It’s for creating connections with services that need to be centralized- 3 reasons. One centralized and correct and lightweight does not need containers
It’s for calling things like databricks etc without needing a container
Because of which it can scale high and need very low resources
Also lifecycle like abort on failure can be easily associated