We would like to use Alexa with a custom built and already trained classifier in our company along with some other Alexa intents built for some other purpose (So, Alexa can't be used as a passthrough to the other classifier). As currently it is not possible to capture the entire user utterance directly in the skill backend, we are planning to implement the following solution -
We would add a new intent in the skill named say XXX and train it to capture everything said by the users after a phrase in a "SearchQuery" type slot...
Example - Alexa, ask XXX to check with ABC <searchQuery slot>
And then we send the slot value captured to the other classifier. The whole idea is to be able to use a system which is already built and trained over the years to work with Alexa and avoid the situation where we have to duplicate and synchronise the intents/utterances in two different systems(the other system will still be used in a standalone mode).
Could this be an issue during the certification process? What are the possible alternatives?