When would you use Lex over without using QnABot

QnABot provides a quick and easy way to create a bot for frequently asked questions that can be deployed across many channels. It is built on top of Lex, so if you want to you can build standard Lex intents into the underlying bot (QnABot is Lex under the hood, so you are not painting yourself into a corner). If you are not looking to answer frequently asked questions and just want to build intents for specific self service intents, then a standard Lex bot might be the way to go. Lex has many features that are not available in the QnABot designer. You would have to use the standard Amazon Lex to make use of them. One example might be if you are building a bot for a self service intent meant to be integrated into a contact center using voice. You might want to make use of custom vocabularies to increase the speech recognition accuracy (https://docs.aws.amazon.com/lexv2/latest/dg/vocab.html). You might want to use "wait and continue" to allow a customer time to find their account number (https://docs.aws.amazon.com/lexv2/latest/dg/wait-and-continue.html) You might want to use runtime hints to improve accuracy when collecting a product name or a mother's maiden name (https://docs.aws.amazon.com/lexv2/latest/dg/using-hints.html) You might want to use "conditional branching" to branch the conversation based on a session attribute or a slot value (https://docs.aws.amazon.com/lexv2/latest/dg/paths-branching.html)