Hi Team, 

Currently client is using a smart response setup, wherein they have their own defined templates. 
There's are ML models setup in a Conv AI App for each kind of intent which are mapped to templates in the FAQ Bot. 

This setup is failing to retain the smart responses on a live chat case, as the user message frequency within short time is higher than other channels. 
So the intent that was detected and the smart response suggested go away as soon as user types and extra message, like thanks, ok etc.

What they need is to either:
1) A way to retain the smart response by ignoring the messages where no intent is detected.
2) In case #1 is not possible, any other ways we can suggest like showing some KB articles/Folders based on intent of the message and retain them.

They have also highlighted Sprinklr Smart Response won't be a possiblility for them and they need control on smart response based on the defined templates only they have set.

The idea to bring the AHT down.

Thanks for taking a look at it and appreciate if we can help the client with their use case.