No more slow first request
A keep-warm schedule removes the 8-10 second cold start the API used to have after a quiet spell.
The API runs on-demand and used to take 8-10 seconds to answer the first request after sitting idle. A lightweight health check every five minutes now keeps an instance ready, so normal traffic doesn't pay that cost. A burst of simultaneous requests can still start extra instances.