0

I would love to hear some advice from someone who has programmed for Alexa and any other virtual assistant. I'm the only one in my team with experience with Alexa and just recently we have sold our first Alexa product. The client wants to know how expensive it would be to make the app also work for Siri/Google Home/Cortana but I have no idea about other virtual assistants and we don't really have time to dive deep into every assistants documentation.

That's where you could help me. Do you have any experience in programming for Alexa and any other mentioned virtual assistant? Maybe you have even tried them all? Which ones are the most similar (from the development point of view of course)? Do other assistance have the same intent/utterance/slots-logic as Alexa? The most interesting thing I want to find out is how much Alexa code we could actually reuse and how much time I would need to spend if I wanted to learn another frameworks granted that I already know Alexa API. Can you share your experiences on this topic?

Thank you very much for any advises.

PS forgot to mention, for Alexa we have made our own webservice endpoint and used Java

Kalina
  • 11
  • 4

1 Answers1

1

Yes, I've developed for Alexa, Actions, Cortana, and Bixby.

All have similar "dialog design" guidance and support. Some names change.

Alexa and Actions are, in my opinion, similar in architecture.

Cortana is different from an architectural point of view, because you can bridge the gap between a state-full application and a cloud serverless design. Doing pure SOA and stateless for complex problems can, in my humble opinion, add tons of complexity and "breaking points".

Bixby is just odd; but if you want Samsung mobile...

From a language stand point, Google is JS/webhooks. Cortana also supports JS/node fully, which is nice. Cortana supports C# (first).

Reusing code... will be a problem. You can reuse design, and possibly patterns, and if you built a RESTful microservice of course you can use it from any platform.

It took me about two weeks to learn each platform to a point I published skills. The hardest part (going from Cortana to Actions) was redesigning to be serverless (event based, like cloud functions/aws lambda).

.02

Micromuncher
  • 903
  • 7
  • 19
  • I also used the Alexa Voice Services JDK and Java. It was painful when I used it due to missing libraries and lack of documentation but it might be better now. I had issues using other Amazon services (i.e. Music) re "permission" to use and auth; also probably better now. Also, the Lambda/NodeJS was much better from a doc and "works as expected" perspective. You should also ask about comparison between TOU... Google and Microsoft let you own your software... – Micromuncher Jun 02 '19 at 17:57
  • Thank you very much for your answer, it cleared quite a bit for us! – Kalina Jun 12 '19 at 07:32
  • we are already using a RESTful microservice and our app is pretty simple in terms of complexity, no auth, no account linking etc. But we do not really want to go serverless since we have our own servers and the client wants us to host all services to save money. It was possible for Alexa, but as I can see it isn't really the case with Actions, since Dialogflow is kinda must have and not free. Do you possibly have any experience with making an app for Actions without Dialogflow? Do you think it's possible/worth the savings? Again thank you very much for your input :) – Kalina Jun 12 '19 at 07:38