Hi, I'm Craig.

Welcome to The Treehouse Show.

The Treehouse Show is our weekly conversation with the Treehouse community.

Have you ever had that feeling where you're ready to start using a language or

some new technology that you've just learned about, but

then you realize I don't know where to start.

Now this is a common enough problem that I thought I'd reach out to our resident

expert teacher Chalkers to get some advice.

Thanks for being on the show Chalkers.

>> It's good to be here.

>> I was thinking, that we could talk about a common problem that students have,

and that is the blank slate problem.

And by that I mean when you get to some technology.

You've gone through it a little bit.

You've gotten your hands a little bit wet with it.

You've done some exploring, but you go to start your own project and there is.

Now what do I do?

Now how do I start?

What do I even think about?

How do I use these tools that are in front of me?

What's your take on the blank slate problem?

>> My take on the problem is that you've just gotta start somewhere.

>> Okay.

>> And getting something working,

in a fashion, is probably the best solution.

So can you install this stuff on your machine, and

get it just running with a demo project?

>> Okay.

What do you mean by demo project?

>> So say for example a lot of these services out there, or

even frameworks normally have a hello world example project.

>> Okay. >> That you can just clone.

Some boiler plate code that you can just run on your own computer.

And it's like, that works.

I'm getting somewhere.

Because I think it's useful to have a working model in

your mind of what the system's doing.

>> Okay.

>> And then you can maybe translate that to the documentation a bit better

later on.

But just start tweaking example code first.

>> Okay, so kind of just grab, and most projects have that.

>> Yeah yeah, yeah, yeah.

I think it's almost needed if you're going to interact with developers in any way,

is to have some example projects, and

sometimes there're these things called kitchen sink projects.

So I've worked in a lot of different mobile environments where they say, hey,

here's a kitchen sink application where it shows you all the different menus,

the different styles of animations, things like that.

So you can see the thing in action, and then you can maybe find that in the code,

and then import that into your project.

>> And then kinda take it and make it your own.

>> Yeah. >> Okay.

>> Exactly. >> Neat, so

you pointed earlier this is IBM Watson,this is what you're

working on here?

>> Yes, so Watson is a set of cognitive services from IBM.

It basically allows you to take input from say human speech or text.

>> Okay.

>> And then get the intent and then assign programming functionality to that intent.

So say for example, in this example I've got the treehouse assistant.

And it says what would you like to know about, so I could say.

What is the definition of HTML?

And Watson responds with the definition of HTML.

Now I didn't program necessarily the specific wording of what is the definition

of HTML, but Watson can extract the intent to get a definition.

And then the coding that I provide can

decorate the conversation with these meanings.

>> So it's breaking down the English of that sentence,

and finding out what it was you actually meant.

>> Yes.

>> You could ask that something else- >> Exactly, so-

>> In a different way.

>> So you can see here that the intent that Watson found was getDefinition.

>> Okay. >> And with 94% confidence.

>> Okay, and getDefinition is something you wrote?

That's code that you wrote okay?

>> Exactly.

>> Can I see that code?

>> Sure, so.

The way that Watson works is that you can give it definitions yourself or

you can let your code give the definitions.

Or you can give it an HTML document or

a PDF or a JSON file with other content in, and

that's called a discovery service.

So what I first did was getDefinition.

I wrote all this code myself called handleIntent.

So I don't know if you're familiar with middleware in web frameworks.

>> Okay. >> I basically wrote my middleware for

the conversation.

So if it can find the definition with me, then that's what you'll respond here.

If it can't find the definition with the definitions that I have,

it'll go after the discovery service and discover the meaning.

So let's look at what my code looks like first.

I've created this definition resolver and all this does is it

looks in this definitions file, and then it just extracts itself.

Let me show you what the JSON looks like.

It just looks like this.

So it basically finds HTML, find CSS or JavaScript and

then gets the definition and spits that out into the check.

>> Okay.

>> So that's basically what that definition resolver did.

And that's a lot occurred.

>> So this was based on an example project?

>> Yeah.

>> Okay, so how did you do that with Watson?

What's the example project that you looked at?

>> So the example project was Watson

simple chats, conversation.

So here it is.

>> Okay, cool.

It's a GitHub. >> It's a GitHub thing, and

you pull it down and you can check the demo here.

So this example was to augment a car.

So turn on the windscreen wipers or something like that.

Turn on wipers, let's see if this will do it.

I'll turn on the wipers for you.

So, this is the initial project that I fought.

I fought this in and

then I started augmenting this with the treehouse assistant.

>> Cool. >> So previously I've worked on an Alexis

scale that did something very similar.

>> Right, I thought I was getting deja vu there.

Okay so the Alexa Skill does something similar to that.

>> Very similar, it translates the English into an intent,

which is getDefinition, and then we can grab the definition out.

>> And these are actually kind of similar, aren't they?

I think they actually are both called intents.

>> Yes, exactly, so once you've worked with Alexa, it's easy to

translate some of those skills over into Watson and these other services as well.

All of these cognitive services, because if you think about when you speak,

when you say a sentence, there's always an intent behind it.

Or when you ask a question there's an intent derived, so

these cognitive services help derive intent.

So it can,

Also get the sentiment behind things and- >> Like

if you're angry or- >> Yeah.

>> Wow. >> Yes, say for

example- >> That's really neat.

>> You could, [LAUGH], pass all the information through from a chat from

a custom service representative and a customer, and see their relationship,

and see if anything can be improved- >> Right, wow.

>> There and things like that.

>> This conversation may be recorded and sent through Watson.

[LAUGH] >> Exactly,

so this is the project that I branched up off.

>> And so, let me show you what it looks like.

So if you look here there's an app.js file.

And in here there is just this conversation message with a payload.

And it updates the message and it just sends the response back to the client.

So that's all that does.

It just sends a response back to the client, which just happens to be JSON.

So what I thought was instead of

making my code all spaghettified inside this message, is to pass the payload

forward, or the response port forward, and then augment that with my data.

>> Okay.

>> So that, that was the HTML, JavaScript, and CSS.

>> And did you get there because you started following this, and

were like, this is gonna be.

This is not good.

This doesn't feel good to [CROSSTALK] >> Yeah, yes.

Some of the example code is very like, hey, where should my code live?

There's no best practices for this stuff.

It's just, here's an example for you.

And it works.

>> Right, it's kinda to spark idea, right?

>> Yeah.

>> So now, you've worked on two voice-based apps.

Do you have any ideas that came from that?

>> Yeah, definitely.

Yeah, I want to augment the house with Watson, so that.

And whether if it's through voice, through Alexa, you can tie that into Watson or

whether if it's a chat bot that I can just send text messages to or things like that.

I can basically switch off my the lights around my house because all

the intents are switch off my lights or turn them on.

>> Right. >> You can derive the intent from that

speech.

>> Like a butler of sorts.

>> Yeah, exactly.

>> Okay, cool.

>> Yeah.

>> So that butler idea, that came about because you played with this.

You didn't know how these things worked before, and what you could do with it.

>> No, and I think that's the whole point is that,

as soon as you get a model in your mind then you can apply that

to the world around you, or problems that you're having, or you can see.

I'm very into JavaScript and the Internet of things, so

that was a very natural place for me to go with this.

But I can see that many people who start playing around with this sort of stuff

if they're working QA, or if they're working support,

or all these different areas, they can say, well I can augment the experience,

whether if it's a full pledged AI that can help.

Like hey, can you tell me my latest order?

And what's your order number?

And then it can grab the order number and set from an API and

send that back in a chat.

Rather than clicking through the menus of a webpage,

you can more have a conversation about the thing.

>> So you kind of follow this methodology all the time.

You always just dive into the tech, and I am always amazed, what are you doing now?

You must have started this a long time ago, this practice of just diving in.

If you could tell yourself, give yourself advice right when you were first

learning how to code, what would that be?

>> Persevere, the pain that you go through, it always pays off,

because eventually something will click in your mind, that that's how it works.

And literally, you can do anything as long as you persist with it,

and stay at it.

But when I say stay at it I'm not saying you should have a break because

generally a lot of the ha moments is when you go walk away from the computer and

you go, it's that.

>> Yeah.

>> But yeah, definitely that feeling of being uncomfortable that this is hard,

this is difficult, you'll never understand this.

Those type of feelings are normal and it does pay off eventually.

>> All right.

Thanks, excellent advice.

>> Awesome.

>> Thanks for being on the show again, Chalkers.

>> No worries.

Thank you. [SOUND]

>> Check the notes attached to this video

for links to some great example starter projects.

Blank slate be gone.

Thanks for watching the Treehouse Show.

To get in touch with the show, reach out to me on Twitter, or

hit us up in the Treehouse community.

See you next time.