What Everybody Wants and How to Get It!

[jwlv]

I think the biggest thing that most Amazon Echo users want is customizations to what they need. But Amazon is kind of slow on customization. But I have a better idea.

Instead of depending on Amazon to add support for dozens or hundreds of different services, why not have a one-fits-all type of service. In my idea, anything that Echo does not understand, it will go to an alternate service that the user has set up. Maybe one key phrase would bring up one service, another key phrase would go to another service.

For example, I want to say, "Alexa, read my private messages on EchoTalk."  Well, obviously the Echo wouldn't know how to do that. I could set up an alternate service to 3rd party website. In this example, the URL could be something like http://echotalk.org/echoprivatemsg.php?xyzlogincredentials... When Echo hears the phrase "read my private messages on Echotalk", it will go to the URL that I set up and pass along some parameters such as the login info. The website will display my private messages and Echo will read them to me aloud.

Another example. I ask, "Alexa, is my garage door open?"  I want Echo to go out to my private website, (http://example.com/mygaragescript.php) and my Raspberry Pi unit in the garage would respond with Yes or No. Echo will read that aloud.

As a teacher, I could say to the remote, "Alexa, mark John Smith absent today" and it could go to a private URL where I can write a customized program to hook into attendance area in Infinite Campus. The program would look for a student named "John Smith" in the roster and mark him absent. If successful, the program will display "OK, done." If something went wrong, it would display "Error" and the Echo would read that response to me.  I'm sure the district won't allow me to do this, but it's just an example of the endless possibilities.

This type of open integration will allow users far more flexibility on what they can do. When a user decides to use an alternate service, the response would be given back to Amazon Echo in plain text to read via text-to-speech. It doesn't require Amazon or Echo to use too much of its own resources. It just needs to recognize the key phrases, go out to the appropriate 3rd party service, and read the results from that service.

This could work for a huge variety of services. Whether it is for reading books, RSS/websites, customizing home automation, and whole lot more. All it needs is for Amazon to open some kind of URL, wait a few seconds, and read what come back from that URL.

Perhaps if this was made available soon, the next step is to allow streaming content from a 3rd party website. Instead of reading back text, it could stream back specific types of audio.

There are so many possibilities if this kind feature was available.

Just a thought!

[ScarlettPhoenix]

I might be showing my stupidity here but isn't what you're talking about called apps?
I would download an app for "Gradebook" if I was a teacher that would help me with the things you suggested.
There could be an app (I'm thinking IMDB) Movie times in my area/my favorite theatre
A kind of Yelp app for 'my favorite restaurants'?

Third party people really need to get on board the Echo train.

[jwlv]

Yes, it would function just like an app, but 1000 times simpler.

Right now, Echo has no apps. You cannot put any apps on the Echo and you cannot remove any functionality from the Echo. You have to rely on Amazon to create what they think you want.

Whether you are thinking Apple IOS, Android, or Echo, an app is usually downloaded to your device and it lives on your device until you decide to remove it. What I'm thinking in my idea is not an app in that sense, although it can work just like one. You don't download anything to your device, it doesn't live on your device. It's much simpler because Amazon does not have to develop an environment for apps. It's just a simple key phrase that actives one function. That one function is to go out to a URL and read what it says. The app part of it lives on some website that either you made or some 3rd party made. Because it doesn't live on the Echo or on Amazon's servers, security is a non-issue. The so called app doesn't live on your Echo and has no way to access any info from your Echo or your Amazon account. On the other hand, if you install an app to your Echo, there could be some potential security risks.  Since your login info is stored somewhere on your Echo and apps also live in your Echo, some information can potentially be "stolen" by an app.

An analogy of this is a web browser. A web browser can go to any website you type in and show you lots and lots of information. If you want movie info, you can type in imdb.com. If you want restaurant reviews, you can go to yelp.com. (Keep in mind that not every website has an app that goes along with it). A web browser can be used for millions of things and you don't have to install any apps. Sure, your iPhone or Android both have apps for IMDB and Yelp but you can get the same information by using just a web browser. My idea makes this entire system so simple. You don't need apps. All you need is a way to go out to get the information and have the Echo read it aloud. And all that Amazon needs to do is allow us do to that.

Here's the gotcha. If Amazon decides to go this route, I think it would be a lot harder for Amazon to monetize from it.

[Monty]

Amazon seems to be keeping a tight reign on developers being approved/accepted. I can understand that since the product still seems to be in beta mode.

IFTTT helps somewhat, but so far I have found it useful only as a 'workaround' to get something done.

Given the control Amazon has over this (kinda like Apple does in the iStuff world, LoL) system, I suspect any efforts to offload commands, sensor monitoring, etc will remain as 'workarounds' and not tightly integrated. But, who knows; where there is a will (and money) there is often a way!

Monty

[LoveMyEcho]

jwlv - Something like what you describe is already in the works, by another Echo dev with whom I'm acquainted. I'm not sure it's okay to share details at this early stage, but I can tell you this: he's developing a cloud-based service that can interact with pretty much ANY cloud-connected device with an Application Programming Interface (API).

Yup: we're talking everything from car computers to the chip in smart coffee makers.

The dev is in close contact with Amazon's own engineers, so I'm very confident this is happening. Stay tuned!   

[fstbusa]

any update on this?

[Monty]

After using the Echo for about 10 months, it has exceeded my expectations! But, like many of you, I want it to handle "my things/stuff/commands." In other words, personal service!

This alternate server idea is great, and I see a number of projects out there where you can use a Raspberry Pi or other PC, or even a virtual machine to do these custom tasks.

However, while I would have jumped on these programming challenges 40 years ago(!), I simply haven't the stamina at this age to jump back into the coding thing and learn new languages so far removed from FORTRAN (yikes!) of my early years!

I have looked at coding skills at Amazon's developer sites, and while they make it as easy as possible, it is still a daunting learning curve. And the other projects also require coding/programming knowledge equally as daunting if you want to "customize" the Echo to your specific needs and not just following the bouncing ball of some set activity or function.

The challenge as I see it is just how many specific "alert words/phrases" you have to remember to make use of even the current Alexa skills. Each skill requires a unique alert word or phrase so Alexa can get you to the right function. And just how many of those unique words/phrase can an average user remember?

As an example, I wanted the Echo mainly for home automation and home entertainment system control. As in a fair amount of lights, door/window sensors, motion sensors, along with control of a fairly elaborate set of Logitech Harmony Ultimate control functions. And how about that Campbell Soup recipe thing...is it "Campbell Recipes", "Campbell Soup Recipe", or "Campbell Kitchen Recipes"? I forgot as I have not used it for a week or so!

But even though i use the light on/off commands daily, I easily forget the exact phrase I have set in Alexa to turn on a specific light or group of lights. Even tho I make the name as user friendly and associative as possible, I still often get one word wrong, ie is it "Basement light 1, 2, or 3"? Or do I just give up and say "Basement Group", wasting electricity because I can't remember exactly which room or area I put which numbered light in?

What I am getting to is that the "natural language" ability of Alexa and other voice recognition technologies are still at the very early stages. As humans, we develop an intuitive language ability that interprets the 'gaps' and fills in details to arrive at what the speaker intended; far more than artificial language detection can.

Alexa is a computer and you have to be very specific with clear enunciation or you start to get the Alexa version of a shrug of the shoulders, LoL.

So my "want" is simply more advanced language AI, not just more cloud services with the same problems or learning more "alert words/phrases" that currently exist with natural language commands. :-)

Monty

[jwlv]

In the early 90's, I did a research paper on Natural Language Processing (NLP) and Neural Networks while in college. Some 20 years later, not much has changed. There are very few people who are active in the Artificial Intelligence field when compared to other fields in technology and computing.

Whether you ask Siri, Alexa, Google, or Cortana, what you ask is as important as how you ask.
Our brains can piece together information and even insert missing information or discard extraneous information so that we can comprehend a statement or question.
To us, these questions are equivalent and the answer is the same:
1. How many inches is five feet?
2. Five feet is how many inches?
3. How many inches is a five foot pole?

Most NLP's can understand #1 and #2 and give the correct answer of 60 inches. By changing the question slightly, NLP's get completely dumbfounded and unable to process the question being asked.