@veejae
That makes sense.
Frankly said, It has worthy of making gAction instead simple transcriptionHook (But no matter what, you can use transcriptionHook also.)
Because, your visitors(I guess) can ask anything, so word itself is not important, intention is matter.
By example; For the company history, you probably consider this voice question. “Tell me history of this company”. Maybe you think it could be done by just hooking “history” word or whole sentence “Tell me history of this company”.
But your guests may ask like these. “About this company, Tell us whatever.”, “How old this company is?”, “Could you inform us about background?”… So all these possible transcription hookings are impossible. But you can train AI with sample sentences to catch the intention of user with gAction.
The problem is, making proper gAction is really not so easy. At least, you need to understand JSON format and basic concept of google action for Assistant.
About the response; Usually 300 characters as voice response is recommended (max 600 chars). For longer response you have two options. 1) Use TTS solution to read long text on real time. 2) prepare your response as mp3/wav file then play it when requested.
I think second option has more sense.
