Order amongst chaos: automation best practice in a serverless world

Without straying too far into any notion of 'cookie cutter' computing, how intelligently can we apply custom-built pre-templated automation controls into serverless environments based upon defined (and agreed) best practices for information management and application deployment?