The "Alexa" name is starting to get confusing. Amazon already owns something called Alexa, a website/service which provide information about a website's stats and ranking.
Worse, Amazon also offers an API/SDK to Alexa via AWS [1], which offers a surprising amount of information about a website that they gather (Global Rankings, regional rankings, standard <META> tag scrapping, DMOZ info, IP/location info, email address/telephone/address and other contact info, bunch more stuff). It's actually pretty amazing, but the API and documentation is terribly out-of-date.
This got my hopes up that Amazon had done an overhaul of the Alexa rankings/info API... [sigh]
I thought "Alexa" was their attempt at general search, which gave stats among other things. In that sense, Echo's usage was natural, albeit confusing in terms of brand.
Lots of code examples in JS and Java (even though I believe a github repository would have been better than zip archives...), a $100M investment fund, and a real integration with their services and devices. Amazon seems to be really serious about this.
When Siri came out, I found myself wanting to build a service using it and got disappointed (even though the reasons not to open the API were obvious), so I'm now happy to see Amazon calling developers with a voice based API. I'd like to see how this will pan out.
Exited to see these new SDKs for the Echo, using it daily it definitely gives you a feeling for the vast potential of the platform given 3rd party contribution. Using AWS Lambda also makes development of Echo apps makes it extremely easy, it took me only about an hour to get a working prototype of a simple app working.
I'm looking forward to attending the Boston amazon developer day this weekend to really dig deep into all the sdks and get some questions answered from the team.
AWS Lambda definitely seems like the way to go. It looks like hosting the service yourself gets complicated very quickly when you have to factor in things like authenticating the requests. [1]
I'm also not sure how it would technically work, but I do wish there would be some way to run this locally. I would love to get more access to some of the hardware on my local network, but I would be hesitant to open the same up to the public internet. I guess Amazon is (smartly) targeting more consumer focused developers and not hobbyists working on their own pet projects.
Just sold my Echo after getting little value from it - must admit I was holding out for an API, I think it will be 100x more useful once we see some good 3rd-party integrations.
You are totally right with your point and I upvoted you for that - I wouldn't it reduce to priorities only though. There's desinterest and of course simply the fraction with "no clue" what's going on within the world of big data sales or gov/mil related surveillance.
The best systems--Siri, Google Now--have 95%+ ASR accuracy AND can find the right answer among millions of possibilities once they've broken down the utterance.
Combining human-like ASR & NLP with super accurate search capabilities is hard.
it would be great if there was no namespacing, and multiple developers could register the same phrase. alexa/echo (what's this thing called?) should learn based on your usage what service you want to respond to a certain phrase.
For example, "turn on the lights" should work with whatever smart home hub alexa can find on it's wifi network, not just with whatever smart home company registers the phrase first.
I don't believe they will need to be unique. Users can browse for skills and manually enable them. I'm sure the app will alert you that there's a conflict when you attempt to enable a new skill that uses the same invocation name.
> Amazon is investing up to $100 million to support developers, manufacturers, and start-ups of all sizes who are creating unique and innovative experiences designed around the human voice. Whether that’s creating new Alexa capabilities with the Alexa Skills Kit, building devices that use Alexa for novel voice experiences using the Alexa Voice Service, or something else entirely, if you have a visionary idea, we’d love to talk to you.
I just ordered an echo for my hobby workshop, i'm super excited for this thing. My dream is to be able to import google sketchup models (somewhere) than as i'm working be able to ask it questions about the model.
So like, imagine i'm standing in front of a freshly milled piece of oak, and i'm roughly marking out my pieces:
"Alexa, What are the dimensions for the top side apron"
With Alexa, you have to bet on Amazon winning the IoT game. Amazon is great at ecommerce (mostly because they don't care about making money), but it's not clear they're great at platforms.
But if all you want is the functionality of an Echo that you can customize, then go with a well-funded startup staffed by some of the best AI engineers in the world offering similar and often better capabilities, and build your own so you can integrate it wherever you want.
34 comments
[ 4.7 ms ] story [ 79.7 ms ] threadWorse, Amazon also offers an API/SDK to Alexa via AWS [1], which offers a surprising amount of information about a website that they gather (Global Rankings, regional rankings, standard <META> tag scrapping, DMOZ info, IP/location info, email address/telephone/address and other contact info, bunch more stuff). It's actually pretty amazing, but the API and documentation is terribly out-of-date.
This got my hopes up that Amazon had done an overhaul of the Alexa rankings/info API... [sigh]
[1] http://aws.amazon.com/awis/
When Siri came out, I found myself wanting to build a service using it and got disappointed (even though the reasons not to open the API were obvious), so I'm now happy to see Amazon calling developers with a voice based API. I'd like to see how this will pan out.
https://github.com/AreYouFreeBusy/AlexaAppKit.NET
I'm looking forward to attending the Boston amazon developer day this weekend to really dig deep into all the sdks and get some questions answered from the team.
I'm also not sure how it would technically work, but I do wish there would be some way to run this locally. I would love to get more access to some of the hardware on my local network, but I would be hesitant to open the same up to the public internet. I guess Amazon is (smartly) targeting more consumer focused developers and not hobbyists working on their own pet projects.
[1] - https://developer.amazon.com/public/solutions/alexa/alexa-sk...
Not for me or my family.
The best systems--Siri, Google Now--have 95%+ ASR accuracy AND can find the right answer among millions of possibilities once they've broken down the utterance.
Combining human-like ASR & NLP with super accurate search capabilities is hard.
Do they need to be unique across all skills?
Will there be a land-rush on all the good two and three work invocation names?
Do I manually select which skills will be enabled for my Echo, and only those invocation names will be accepted? What about conflicts?
For example, "turn on the lights" should work with whatever smart home hub alexa can find on it's wifi network, not just with whatever smart home company registers the phrase first.
So like, imagine i'm standing in front of a freshly milled piece of oak, and i'm roughly marking out my pieces:
"Alexa, What are the dimensions for the top side apron"
Even cooler, if i had a camera in there too
"Alexa, where did I put my tape measure?"
With Alexa, you have to bet on Amazon winning the IoT game. Amazon is great at ecommerce (mostly because they don't care about making money), but it's not clear they're great at platforms.
But if all you want is the functionality of an Echo that you can customize, then go with a well-funded startup staffed by some of the best AI engineers in the world offering similar and often better capabilities, and build your own so you can integrate it wherever you want.
That's MindMeld.