Hacker News Reader

I'm curious, what's the difference you're seeing between those three platforms for your use case?

The biggest is invocations – as more skills crowd the platform, getting Alexa to recognize our invocation is getting unusable.

Training for our specific pronunciation isn't available on Alexa. Our app is called Tali, pronounced "tally", and to get it to work now you have to say "Tay-lie". Whereas Google Actions can be trained with custom pronunciations.

I should also say that Dialogflow is magnificent.

Have you written up your experiences in a blog post or talk or something?

I'm on the same boat right now, was asked to build a POC in Amazon Lex, an Alexa Skill, Microsoft LUIS, Wit.ai and Dialogflow.

So far I've gotten to LUIS and Lex/Alexa, I'm glad to read Dialogflow is better :) (had high hopes for it)

You can add custom pronunciations on your amazon skills too

Reply on news.ycombinator.com