Basically it means that with Kotless engine for Ktor you will be able to deploy Ktor to AWS as easy as a native Kotless app. Kotless introspects code, get the information about routing and static resources and create the deployment from Ktor application code.
Also engine is optimized for use in cloud -- it does not include any unnecessary for Serverless execution web-servers or anything else. And yeah, no reflection is used 🙂