WEBVTT

00:00.280 --> 00:00.720
Well.

00:00.720 --> 00:01.160
Hello.

00:01.200 --> 00:02.400
My name's Ed Donner.

00:02.440 --> 00:06.880
I'm a two time AI startup co-founder and a former managing director for J.P. Morgan.

00:06.880 --> 00:10.240
And I'm about to take you on a wild ride.

00:10.280 --> 00:13.920
We're going to be spending the next six weeks together on an adventure.

00:14.000 --> 00:17.800
I've actually just finished myself doing the six weeks worth of projects.

00:17.800 --> 00:23.000
There are eight projects along the way, and I can tell you I have some astonishing things in store

00:23.000 --> 00:24.760
for you and I can't wait to get started.

00:24.760 --> 00:25.600
Let's do this!

00:26.200 --> 00:30.560
Now, it's customary with these things to kick off with an introduction to myself and talk about the

00:30.600 --> 00:35.400
goals of the course, but people who've taken my courses before will know that I don't like doing that.

00:35.440 --> 00:36.360
Let's not do that.

00:36.360 --> 00:38.200
Let's get straight to action.

00:38.200 --> 00:44.200
Let's go and see some autonomous AI agents in practice, and we'll come back and do that other stuff

00:44.200 --> 00:44.640
later.

00:45.560 --> 00:51.720
So I've taken us now to the app N810, which is an exciting app that is definitely generating interest.

00:51.720 --> 00:58.680
This is an example of one of these low code or no code workflow apps that allows you to construct a

00:58.680 --> 01:04.090
workflow of orchestrating between different applications and build it without necessarily needing to

01:04.090 --> 01:05.410
write any code yourself.

01:05.730 --> 01:11.650
The difference with some of the other workflow tools like this is that it has generative AI built into

01:11.650 --> 01:15.330
it, and that's why people are so excited about N810.

01:15.610 --> 01:21.530
So if you come to N810, which is the cloud version of it, you can also download it and run it locally.

01:21.530 --> 01:23.210
But here we're using it in the cloud.

01:23.210 --> 01:26.850
You can press Get Started and set up an account if you want to do this yourself.

01:26.850 --> 01:28.010
Or you can just watch me.

01:28.010 --> 01:31.170
But it might be fun to try this out yourself if you haven't used it before.

01:31.490 --> 01:37.050
So I've set up an account, I've done that, and once I've signed in, I get this beginning screen to

01:37.090 --> 01:38.810
add my first step.

01:38.810 --> 01:42.490
You can also use one of their templates right off the bat if you want to, but we're going to build

01:42.490 --> 01:44.210
something ourselves right now.

01:44.490 --> 01:46.170
So we're going to add a first step.

01:46.170 --> 01:51.010
And what we're going to do is we're going to begin by starting with a chat message.

01:51.290 --> 01:57.130
So this if I go back to the canvas here, you press back to canvas on the top left.

01:57.130 --> 02:03.130
We can now see this canvas which is showing that we have a chat box here that's represented by this

02:03.130 --> 02:04.770
chat window down here.

02:05.070 --> 02:09.030
and we can chat and do something as a result of that chat.

02:09.070 --> 02:10.430
So what do we want to do?

02:10.870 --> 02:16.430
Well, we press the plus button right here and we come over to the right and we decide that we want

02:16.430 --> 02:18.710
to have advanced AI.

02:19.270 --> 02:22.950
And we're going to create an AI agent right here.

02:23.150 --> 02:28.230
And now, if I go back to canvas, I'll show you that what we have in our canvas now is a chat message

02:28.230 --> 02:31.430
going into an AI agent representing by this box.

02:31.750 --> 02:34.350
And then something can come out at the back of that.

02:34.630 --> 02:36.190
And you don't need to know all of the details here.

02:36.190 --> 02:37.550
This is more to give you an idea.

02:37.590 --> 02:39.270
And by all means you can play with this yourself.

02:39.270 --> 02:42.110
But most of this course is going to be about building things like this.

02:42.430 --> 02:43.910
But let's give this a shot.

02:43.910 --> 02:46.990
So the next thing we can do is add a chat model.

02:46.990 --> 02:50.870
And this is a large language model that's going to sit behind this AI agent.

02:51.030 --> 02:54.270
And I'm going to pick open AI chat model.

02:54.390 --> 02:58.270
Now that's because I have an open AI key which will be setting up for you if you don't have one already

02:58.270 --> 02:58.990
quite soon.

02:59.030 --> 03:04.790
And alternatively, if you don't want to do this, you can use Olama instead and run for free, but

03:04.790 --> 03:10.960
you would then have to install it Locally, but I've set up my OpenAI account and you can do this yourself.

03:11.120 --> 03:16.800
All you do is you come down here and you press create New credential and you type in your API key.

03:16.920 --> 03:18.000
But I've done that already.

03:18.000 --> 03:20.440
So my OpenAI account is linked.

03:20.480 --> 03:21.720
It's as simple as that.

03:21.760 --> 03:22.760
And here it is.

03:22.800 --> 03:28.280
It looks here OpenAI chat is connected to my chat model.

03:28.280 --> 03:29.280
So far so good.

03:29.480 --> 03:30.680
Let's keep moving.

03:30.720 --> 03:32.680
We're now going to add in a tool.

03:32.800 --> 03:38.160
Now, tools are one of the most fundamental building blocks of a gigantic AI.

03:38.360 --> 03:41.120
And this week is going to be a lot about building tools.

03:41.520 --> 03:46.160
So I pressed the plus button to open up the list of tools that I could connect with the AI agent.

03:46.160 --> 03:50.520
And you can see there's a large number of tools, a lot of different things that this agent could,

03:50.520 --> 03:52.560
could use, could take advantage of.

03:52.600 --> 03:54.920
It could do calculations with a calculator.

03:54.920 --> 04:01.640
It can book meetings using my Google calendar or read my emails, look at my Facebook timeline, lots

04:01.640 --> 04:04.600
of things that it can do, as you will see here, read Hacker News for me.

04:04.600 --> 04:05.800
I would like that.

04:05.800 --> 04:10.220
Okay, but we're going to go all the way down and we're going to find something here that I just want

04:10.260 --> 04:13.620
to show off, which is the Philips Hue tool.

04:13.700 --> 04:15.140
I don't know if you know about Philips Hue.

04:15.180 --> 04:17.660
It's something which is, uh, gives you light bulbs.

04:17.660 --> 04:22.620
And I have a lot of light bulbs in my apartment here, and we're just going to pick a light bulb.

04:22.620 --> 04:24.660
This one here called the bed strip.

04:24.900 --> 04:26.580
And we can add a field.

04:26.580 --> 04:27.900
We'll say brightness.

04:28.700 --> 04:33.140
We'll let the AI automatically choose what it wants to do with brightness.

04:33.340 --> 04:36.540
We'll add a field hue which is like the color.

04:36.780 --> 04:38.460
And we'll let that do that too.

04:38.460 --> 04:39.740
And that's all there is to it.

04:39.780 --> 04:45.700
We've now when I go back to the canvas, you'll see that we have given a tool to this agent that you

04:45.700 --> 04:50.860
can see visually here, that it is able to control the Philips Hue light bulb.

04:51.260 --> 04:56.860
So here we are looking at our chat, our agent, the LM and the tool.

04:56.900 --> 05:01.300
Let's get going with bringing up the chat message that's associated with this input.

05:01.460 --> 05:04.020
And let's say well hi there.

05:04.740 --> 05:08.220
And you can see it working and thinking hello, how may I assist you today.

05:08.420 --> 05:13.150
Let's say please Turn the lights on.

05:13.550 --> 05:15.350
Bright white.

05:16.630 --> 05:17.230
And let's see.

05:17.270 --> 05:18.270
And bam!

05:18.790 --> 05:19.870
Did you see that coming?

05:20.110 --> 05:24.230
On come the lights that I am, in fact, wearing my light strip.

05:24.510 --> 05:25.990
Uh, so here we go.

05:26.150 --> 05:28.150
Now, you may say to me, okay, that's cool.

05:28.150 --> 05:29.870
I see tools in action.

05:29.870 --> 05:31.950
I see an LM that I'm chatting with.

05:31.990 --> 05:34.150
It's controlling my lights in my apartment.

05:34.190 --> 05:34.910
Great.

05:34.910 --> 05:36.910
But it's not really showing me autonomy.

05:36.910 --> 05:39.830
I'm not seeing something where it's making its own decision.

05:39.830 --> 05:41.510
Well, then let me show you that.

05:41.510 --> 05:46.350
So to make it autonomous, we need to give this agent some agency.

05:46.350 --> 05:49.630
So let's say please pick a color.

05:50.590 --> 06:00.190
Let's say, uh, either deep red or deep blue and change the lights to that color.

06:01.030 --> 06:02.430
Let's see how it does with that.

06:03.030 --> 06:05.910
And there we go a lovely red comes out.

06:05.950 --> 06:07.230
Fantastic.

06:07.230 --> 06:08.590
So there you have it.

06:08.630 --> 06:10.470
We have an autonomous agent.

06:10.510 --> 06:16.330
Maybe it's a simple task, but nonetheless, you just saw it making its own decision, picking red over

06:16.330 --> 06:21.610
blue and changing the light bulb all through Nathan's web app.

06:22.370 --> 06:27.970
So look, that was a bit small and superficial, but nonetheless, it was our first foray into the world

06:27.970 --> 06:29.010
of Agentic AI.

06:29.290 --> 06:32.770
And I strongly encourage you to go and take a look at N810.

06:32.890 --> 06:34.650
It's so easy to use.

06:34.650 --> 06:40.610
It's so quick to set up an account, to hook up a web form to your Google Calendar, and lots of other

06:40.610 --> 06:46.690
things to try out, and it gives you a nice hands on sense of what it's like to work with agents.

06:46.930 --> 06:52.970
But I should say, this is the last time that we'll be using a tool, a low code, no code tool like

06:52.970 --> 06:53.250
this.

06:53.250 --> 06:58.210
As a user of AI, for the next six weeks, we're going to be rolling up our sleeves and we're going

06:58.210 --> 06:59.810
to be coding agents.

06:59.810 --> 07:05.570
We're going to be building them ourselves and working with frameworks that allow us to build and have

07:05.570 --> 07:10.250
different agents interact, be orchestrated to solve problems.

07:10.730 --> 07:11.290
Okay.

07:11.570 --> 07:14.890
So in the next lecture, I will tell you more about what I have in store for you.

07:14.930 --> 07:15.570
See you then.
