No problem, I'll provide an example:
Suppose I have a GET method endpoint which receives and query parameter.
Based on said query param it run logic, access DB, storage etc. and produces a response.
(Very) Basic illustration:
GET /process?q=1 ==> HTTP 200 OK "response for q=1 is ...."
Now, as the response in the same for every given query param (when q=1 it always produces same result etc.), I don't want to calculate the response each time by invoking a lambda function. I'd like the API gateway to issue a response from cache. Thus reducing latency and providing better performance.
Also. I'd like to configure caching properties for each endpoint specifically and not in the AWS console but in serverless.yml.
Hope It helps to clarify the issue.
you can see an article discussing these features in version 0.x here