WEBVTT

NOTE Sentence-level transcript of https://www.youtube.com/watch?v=ZGzthw4BZiA

NOTE One cue per sentence. Cue ids are the line anchors on /transcripts/ZGzthw4BZiA.html. A cue ends where the next begins, or 2 s after its last word.

s1
00:00:12.960 --> 00:00:17.520
Hello everyone and thank you for coming to our workshop building with perception agents.

s2
00:00:17.520 --> 00:00:19.040
Uh my name is Emil Basel.

s3
00:00:19.040 --> 00:00:23.720
I lead a team of solutions architects uh within Amazon AGI Lab.

s4
00:00:23.720 --> 00:00:25.560
And our team is focused on two things.

s5
00:00:25.560 --> 00:00:34.080
The first is working with some of our earliest adopter customers who want to work with some of our early models and products to help them get into production.

s6
00:00:34.080 --> 00:00:46.200
And the second part is closing the feedback loop from our customer engagements, uh building evals for our research team to train our models on, and bringing product uh and engineering feedback

s7
00:00:46.200 --> 00:00:48.120
to those teams as well.

s8
00:00:48.120 --> 00:00:55.480
And I'm joined by one of our stellar solutions architects, Shruti Arora, who's going to be running the workshop today.

s9
00:00:56.040 --> 00:00:58.360
Uh just a bit about what we're going to be talking about today.

s10
00:00:58.360 --> 00:01:06.920
I'll be giving a quick overview of the Amazon AGI Lab, and talking about the first product that we shipped last year, Amazon Nova Act.

s11
00:01:06.920 --> 00:01:14.960
And then we'll get into perception agents, and that's what the focus of today's workshop is, and then we'll get into the workshop.

s12
00:01:16.600 --> 00:01:17.520
So a bit about the lab.

s13
00:01:17.520 --> 00:01:27.360
So we are a frontier research and product team that combines startup speed with the scale and the resources of Amazon.

s14
00:01:27.360 --> 00:01:38.400
And we are really focused on building knowledge work agents that can perceive, reason, and act in a variety of different environments to complete real-world tasks.

s15
00:01:38.400 --> 00:01:41.800
Um our offices are in San Francisco and Boston.

s16
00:01:41.800 --> 00:01:44.440
We've got a few other talks going on throughout the week.

s17
00:01:44.440 --> 00:01:52.600
We have a booth at the expo, so stop by to collect some swag, meet some folks from the team, and hear some of the talks there.

s18
00:01:52.840 --> 00:01:57.600
So I want to share a bit about our approach and how we think about agents.

s19
00:01:58.040 --> 00:01:59.080
Across three different ways.

s20
00:01:59.080 --> 00:02:00.560
So, the first is around collaboration.

s21
00:02:00.560 --> 00:02:05.640
So, we really think about building agents that are teammates much more than just a black box.

s22
00:02:05.640 --> 00:02:17.000
And when we say teammates, we really want an agent that can understand what our goal is, our objective is, and can share how they're thinking about achieving that goal.

s23
00:02:17.000 --> 00:02:25.320
And and then grounding all of that in actual customer uh use cases.

s24
00:02:25.320 --> 00:02:32.800
And so, we work very very closely with a large number of customers to really help ground all the work that we're doing with agents.

s25
00:02:32.800 --> 00:02:35.040
And the second is around perception.

s26
00:02:35.040 --> 00:02:37.880
So, agents need to understand their environment.

s27
00:02:37.880 --> 00:02:44.120
And so, for us, because it's a perception-based agent, the environment spans multiple devices.

s28
00:02:44.120 --> 00:02:53.040
So, we started with browsers last year, but you can imagine any visual screen is an environment for our agent to carry out knowledge work in.

s29
00:02:53.040 --> 00:02:59.120
So, mobile devices, and also computer use is the other one.

s30
00:02:59.120 --> 00:03:01.200
And the third one is reliability.

s31
00:03:01.200 --> 00:03:06.080
And we have a great customer quote that said, "If my agent works 50% of the time, it doesn't work at all."

s32
00:03:06.080 --> 00:03:16.720
And that's because you're either not going to run it at all, or if you do run it, you're going to end up uh debugging it and spending all this time, and you're going to basically do away with all the gains that you're getting from that agent.

s33
00:03:16.720 --> 00:03:23.560
And so, we spend a lot of time building and training our models in uh simulated environments for reinforcement learning.

s34
00:03:23.560 --> 00:03:36.840
Um and again, a lot of those are based on real-world uh use cases, which we get from not only our external customers, but a lot of internal Amazon teams that are carrying out a lot of uh knowledge work.

s35
00:03:38.200 --> 00:03:38.440
Cool.

s36
00:03:38.440 --> 00:03:42.440
So, with that backdrop, uh I want to talk about the first product that we shipped last year.

s37
00:03:42.440 --> 00:03:54.560
So, Amazon Nova Act is a first-class AWS service that allows developers to build and run reliable browser-based workflows at scale.

s38
00:03:54.560 --> 00:04:08.959
So, using natural language, you can drive a browser workflow to populate a browser, navigate, search with it, fill out forms, and call a human in the loop if you need a human for a expense report approval,

s39
00:04:08.959 --> 00:04:13.320
or if you get stuck, or other things like that.

s40
00:04:13.360 --> 00:04:17.600
And there's kind of four primary use cases which we've seen customers adopt this for.

s41
00:04:17.600 --> 00:04:19.440
The first is form filling.

s42
00:04:19.440 --> 00:04:27.040
Uh so, if you have to do that at scale across a large number of different websites, and you don't want to write a prompt for each single website, or write like a

s43
00:04:27.040 --> 00:04:33.960
like a like a a deterministic script for each single site, you can use Nova Act for that.

s44
00:04:33.960 --> 00:04:35.320
Uh search and extract.

s45
00:04:35.320 --> 00:04:42.520
If you want to do research and get data from a lot of different websites, shopping, and QA.

s46
00:04:42.520 --> 00:04:45.080
So, a bit more on what we launched.

s47
00:04:45.080 --> 00:04:48.640
Uh the Let me just get through all these.

s48
00:04:48.640 --> 00:04:51.160
Uh so, it's a first-class AWS service.

s49
00:04:51.160 --> 00:04:55.760
So, if you go into the AWS console, type in Nova Act, you will see Nova Act there.

s50
00:04:55.760 --> 00:04:58.960
And from the console itself, you can see all of the all of the trajectory.

s51
00:04:58.960 --> 00:05:09.360
So, it's a very easy way to see everything which your agent has done, the bounding boxes around all the elements which it's looking at, and it's trying to take actions on there.

s52
00:05:09.360 --> 00:05:16.080
And then behind the scenes, we have a custom computer use model which is doing all of the all of the inference there.

s53
00:05:16.080 --> 00:05:24.480
This year, we launched an MCP server as well as a skill, and you're going to be using those in the workshop, and that's for your agents to be able to use

s54
00:05:24.480 --> 00:05:26.080
Nova Act very easily.

s55
00:05:26.080 --> 00:05:28.120
And then we've got human in the loop and tool use.

s56
00:05:28.120 --> 00:05:29.400
And we also have a playground.

s57
00:05:29.400 --> 00:05:37.360
So, if you go to the Nova Act homepage, it's a no-code way to just see Nova Act in action there.

s58
00:05:37.360 --> 00:05:43.000
Now, looking ahead, and the topic of today's workshop is perception agents.

s59
00:05:43.000 --> 00:05:46.440
Now, why do we feel there's a need for perception agents?

s60
00:05:46.440 --> 00:05:55.400
So, uh for those of you who are not familiar with the term of computer use, uh that's basically the way for an agent to take an action on a computer screen, right?

s61
00:05:55.400 --> 00:06:00.520
So, you see a blue button in the right corner, and you tell your agent, "Hey, go click this blue button over here."

s62
00:06:00.520 --> 00:06:10.920
Very important, very powerful, it's very important for uh automating workflows, but we feel that there's another half of that, which is the agent actually understanding what it's seeing on the screen itself,

s63
00:06:10.920 --> 00:06:18.440
not just using that as a sort of blind uh screen that it can just go and click on elements there.

s64
00:06:18.440 --> 00:06:22.560
And we really want this agent to be able to understand screen just like a human would, right?

s65
00:06:22.560 --> 00:06:39.720
So, if I showed someone a uh email client from like the late '90s, that was like a Yahoo email client or Hotmail client or something which you've never seen before, you would probably be able to figure out how to send an email from that, even though you've never used that client before, cuz we've

s66
00:06:39.720 --> 00:06:51.440
got this collective understanding of thousands and thousands of hours of using these screens that we kind of intuitively understand how to use these things, and that's really where we're trying to get to with these

s67
00:06:51.440 --> 00:06:53.800
with these agents.

s68
00:06:53.800 --> 00:06:58.400
Uh and the second one is these agents don't require APIs in order to work, right?

s69
00:06:58.400 --> 00:07:07.480
So, if you think about a lot of knowledge work, um it's a lot of humans using screens that just don't have APIs, and that's why there are people who are doing these things.

s70
00:07:07.480 --> 00:07:13.960
And we want these agents to be able to carry out the same work across a variety of sites as well.

s71
00:07:13.960 --> 00:07:19.520
And we want the models to see the screens the same way that we do, because that really facilitates the collaboration, right?

s72
00:07:19.520 --> 00:07:30.040
So, rather than typing up paragraphs for what you want the agent to do uh on this screen and describing the screen, uh you can just point, click, and describe

s73
00:07:30.040 --> 00:07:32.600
changes that you want, and that's what you're going to do in the workshop today.

s74
00:07:32.600 --> 00:07:36.800
It's you're going to click on a couple elements and say, "Hey, I want to change the background to red."

s75
00:07:36.800 --> 00:07:37.360
And that's it.

s76
00:07:37.360 --> 00:07:41.800
And then you just have the agent go go and do that.

s77
00:07:43.240 --> 00:07:53.160
And uh to to start with we've open-sourced two perception agent primitives, and you're going to be using these in depth in the workshop today.

s78
00:07:53.160 --> 00:07:55.160
Uh the first is visual annotation.

s79
00:07:55.160 --> 00:08:02.000
And so what this lets you do is on any webpage, you can point, click, and just describe changes that you want to make.

s80
00:08:02.000 --> 00:08:10.120
And when you're done with that, you click export and it exports out a JSON file with all the changes that you want and a prompt, which you can give to your coding agent

s81
00:08:10.120 --> 00:08:14.800
along with that JSON file to say, "Hey, go and make these changes and do that."

s82
00:08:14.800 --> 00:08:17.800
So important is uh to to note, this has nothing to do with inference.

s83
00:08:17.800 --> 00:08:19.520
This is purely deterministic.

s84
00:08:19.520 --> 00:08:25.560
It just outputs JSON, doesn't have It's not tied to any model or anything else like that.

s85
00:08:25.560 --> 00:08:30.480
The second part uh the second primitive which we've launched is visual verification.

s86
00:08:30.480 --> 00:08:39.080
And if you think about a a site that you have, you've got certain CSS styles and designs which you want your your site to adhere to.

s87
00:08:39.080 --> 00:08:41.919
Uh you've got different workflows which you want your site to support.

s88
00:08:41.919 --> 00:08:45.680
Maybe have like a checkout flow or a login flow or something else.

s89
00:08:45.680 --> 00:08:48.800
And you can describe those as Gherkin files.

s90
00:08:48.800 --> 00:09:00.560
And so you can have visual verification, this this skill, um run by your coding agent against these design artifacts that you have, against these Gherkin files, and make sure that your site is adhering

s91
00:09:00.560 --> 00:09:01.200
to those.

s92
00:09:01.200 --> 00:09:05.080
And again, this is something which you're going to use uh in the workshop today.

s93
00:09:05.080 --> 00:09:08.360
And again, these are all open-sourced, available today.

s94
00:09:08.360 --> 00:09:12.320
And uh yeah, that's what we're going to do in the workshop.

s95
00:09:12.320 --> 00:09:20.280
So I think at this point I've spoken enough, and I am going to hand it off to Shruti, who is going to kick us off on the workshop.

s96
00:09:23.720 --> 00:09:24.920
Do you just press that?

s97
00:09:24.920 --> 00:09:27.080
Thank you.

s98
00:09:27.160 --> 00:09:28.040
All right.

s99
00:09:28.040 --> 00:09:29.320
Can you all hear me okay?

s100
00:09:29.320 --> 00:09:29.839
Yes.

s101
00:09:29.839 --> 00:09:31.120
Okay.

s102
00:09:31.120 --> 00:09:32.320
Welcome.

s103
00:09:32.320 --> 00:09:44.960
Uh thanks Emil for that wonderful introduction to our uh Amazon AGI labs and the products that we launched, we released earlier in May this this year.

s104
00:09:44.960 --> 00:09:59.040
So, as he mentioned, we are going to be doing hands-on workshop, but hands-on on like whatever we released the UI and the visual annotator and the visual verification.

s105
00:09:59.760 --> 00:10:08.840
I have a short demo here, but before we dive into the demo, I just want all of you to like step into this role play here.

s106
00:10:08.840 --> 00:10:11.560
Just just bear with me.

s107
00:10:11.560 --> 00:10:18.200
Imagine you are a front-end engineer and you are building a web application.

s108
00:10:18.200 --> 00:10:23.400
You have a design team and you are taking inspirations.

s109
00:10:23.400 --> 00:10:27.240
They have these Figma files that you are building against.

s110
00:10:27.240 --> 00:10:37.320
You are also collaborating with a sister team that they want to put in a piece of their information on the web page that you are building.

s111
00:10:37.320 --> 00:10:47.839
So, if you are if you think about it that you have the designs, you'll probably take like four to five weeks to like build the designs and build everything code base and

s112
00:10:47.839 --> 00:10:50.520
actually build that website.

s113
00:10:50.520 --> 00:11:03.440
Now, in the third week or like fourth week around that time of your building process, you actually think, "Oh, maybe I should like go back and like verify if this looks fine, if everything looks okay with the rest of the team."

s114
00:11:03.440 --> 00:11:05.360
And you are constantly collaborating, right?

s115
00:11:05.360 --> 00:11:12.960
So, you are checking if if the work that you have done is something that is that is what you want to ship.

s116
00:11:12.960 --> 00:11:28.560
Um so, you send your code base to you share the GitHub or however you are doing your GitLab or whatever you have your code in and you they run that this this web server locally and they open up the website

s117
00:11:28.560 --> 00:11:31.280
and they say, "Oh, this this feels off.

s118
00:11:31.280 --> 00:11:35.240
You know, the head and heading should be probably not in red.

s119
00:11:35.240 --> 00:11:37.720
I should I want to switch it back to something.

s120
00:11:37.720 --> 00:11:40.440
Um, and they provide this feedbacks.

s121
00:11:40.440 --> 00:11:46.920
Probably they will do it in like a Slack message or in a shared doc, for example.

s122
00:11:46.920 --> 00:11:55.560
Uh, and then once you get those feedback on that on your website, you would um, then like go over one of them.

s123
00:11:55.560 --> 00:12:03.520
You'll disagree, agree go to some of them, come to a middle ground, and finalize this is the feedback you want to incorporate.

s124
00:12:03.680 --> 00:12:08.320
You have to go through all of them and one by one then and start making the changes.

s125
00:12:08.320 --> 00:12:19.280
Uh, now with coding agents, I guess it'll take maybe over which used to take maybe a week or it can like reduce to like maybe uh, few few days.

s126
00:12:19.280 --> 00:12:33.000
Now, think about if you had the ability to take their feedback directly and pass it to the coding agent without you your supervision and just like once the changes have been done, that's when you interfere and say, "Oh, this looks good.

s127
00:12:33.000 --> 00:12:35.280
All the changes have been incorporated."

s128
00:12:35.280 --> 00:12:37.480
How would that reduce the time?

s129
00:12:37.480 --> 00:12:38.800
Maybe few hours.

s130
00:12:38.800 --> 00:12:48.040
You just start uh, collect the feedbacks, send it to your agent, and go for a coffee break, come back, and all the changes have been merged.

s131
00:12:48.040 --> 00:12:52.360
So, this is exactly what uh, Emil was mentioning.

s132
00:12:52.360 --> 00:12:57.920
Uh, that we have a Chrome extension that we are calling as visual annotator.

s133
00:12:57.920 --> 00:13:00.000
Um, there are three ways to annotate.

s134
00:13:00.000 --> 00:13:10.680
You can see here, it's either you draw on the page or you pick the element or you drop a pin as you would drop a pin on a Google Maps, say, and provide your annotations.

s135
00:13:10.680 --> 00:13:20.720
And once you have your annotations, you can download these and send it to your team who needs to make the changes or even if it's you who's make the changes who's making the changes,

s136
00:13:20.720 --> 00:13:30.680
then we provide a AI-friendly prompt that you can send it to your agent and it's going to change automatic like well, you have to instruct the agent to like change

s137
00:13:30.680 --> 00:13:33.000
the good base.

s138
00:13:33.600 --> 00:13:36.600
So, this was the visual annotator.

s139
00:13:36.600 --> 00:13:41.960
The next um Oh.

s140
00:13:44.320 --> 00:13:47.840
primitive that we launched is the visual verification skill.

s141
00:13:47.840 --> 00:13:54.720
And when I say skill, it is actually a skill that you can install on Claude or Keynote Keynote power.

s142
00:13:54.720 --> 00:14:05.840
And um you would see these verification as I may mention, these are the design files that have against which you will be running your visual like um verification against.

s143
00:14:05.840 --> 00:14:09.000
Like if does the design colors look fine?

s144
00:14:09.000 --> 00:14:11.360
Is the font size right?

s145
00:14:11.360 --> 00:14:13.160
Is the user flow right?

s146
00:14:13.160 --> 00:14:20.920
But, one thing that will come to your mind like these design specs, these flows, how are these generated?

s147
00:14:20.920 --> 00:14:23.240
How are how how you going to build that?

s148
00:14:23.240 --> 00:14:26.440
Not to worry, we have a solution for that as well.

s149
00:14:26.440 --> 00:14:39.040
When you run your verification for the first time, the there will be a a deep um one level surface crawl that will be done by the UI verification tool

s150
00:14:39.040 --> 00:14:51.080
on your website to go and check what are the links, what are the content, and based on that it's going to create these user flows that are in terms of

s151
00:14:51.080 --> 00:14:59.440
in Gherkin for syntax that the UI UI verification is going to use when it's going to run the verification next.

s152
00:14:59.440 --> 00:15:03.240
So, let me just play that play the demo here.

s153
00:15:03.240 --> 00:15:06.480
Um so, you can see you just like call the tool.

s154
00:15:06.480 --> 00:15:13.000
It's going to go through the design docs specs specs specs and the user flows.

s155
00:15:13.000 --> 00:15:14.560
Uh go through it.

s156
00:15:14.560 --> 00:15:22.880
It's going to open up the browser, run through the design flow user flows, and then generate a report.

s157
00:15:22.880 --> 00:15:28.680
So, this these are the two tools that you will get hands-on in this workshop.

s158
00:15:28.680 --> 00:15:35.520
Uh so, I'm going to jump to what we have how the workshop has been laid out as of now.

s159
00:15:35.520 --> 00:15:38.360
First, as I mentioned, two uh hands-on.

s160
00:15:38.360 --> 00:15:43.520
The second thing that's what we are trying to do is these are the two primitives, of course.

s161
00:15:43.520 --> 00:15:47.000
Perception is not just these two primitives.

s162
00:15:47.000 --> 00:15:51.840
How would you wire these two primitives up for a particular use case?

s163
00:15:51.840 --> 00:15:56.520
There is a scenario that has been uh given in the workshop.

s164
00:15:56.520 --> 00:16:04.360
We are trying to make this fun, so please try it out yourself uh and see like it's like a game day.

s165
00:16:04.360 --> 00:16:09.960
You would see there are like certain hints for the to solve the problem, what the problem is, what the end goal is.

s166
00:16:09.960 --> 00:16:11.960
So, everything is laid out in the workshop.

s167
00:16:11.960 --> 00:16:12.840
Uh it's a GitHub link.

s168
00:16:12.840 --> 00:16:15.320
We will share that with you in a bit.

s169
00:16:15.320 --> 00:16:19.000
Uh so, how will you basically wire these two things up?

s170
00:16:19.000 --> 00:16:25.720
Uh the sec- the next thing we have are um sponsors for the prices.

s171
00:16:25.720 --> 00:16:28.120
We have certain contests that run at the end.

s172
00:16:28.120 --> 00:16:35.839
And um thank you the B team for sponsoring the prices for the uh for the uh event.

s173
00:16:35.839 --> 00:16:54.800
And um we will have a way how you can instead of typing or like annotating on the screen, what are some other ways cool ways of making sure that there are some annotations that you want to do or like feedback that you want to provide about a web page and its designs.

s174
00:16:54.800 --> 00:17:02.120
So, B team will show you how you can do that beyond just typing in your feedback.

s175
00:17:02.120 --> 00:17:10.439
Finally, there are Well, before that, we have After you finish the workshop, we also have like three different options for bonus challenge.

s176
00:17:10.439 --> 00:17:20.959
We will have plenty of time for you to try out these tools in a particular use case that you want to maybe some you want to try wire it up with B

s177
00:17:20.959 --> 00:17:30.800
and see how that how you would probably wire it with any other form of feedback that you would essentially get from your team.

s178
00:17:30.800 --> 00:17:34.280
Or if you think that hey, this is really cool.

s179
00:17:34.280 --> 00:17:39.520
I want to build something that I want to show to my team on Monday like the following Monday.

s180
00:17:39.520 --> 00:17:49.920
So, go for it and show if you're brave enough come in front come at the stage show your use case demo your use case and you'll get a B device.

s181
00:17:49.920 --> 00:18:05.040
The final part is we're going to do a little fun exercise where once you have a good understanding of these two primitives you can think of like multiple ways of how you would use it on a daily

s182
00:18:05.040 --> 00:18:17.360
on your day to day day to day and we will you will submit these ideas of like hey, I think this will be really cool for me to use this and we will have people vote on for these ideas

s183
00:18:17.360 --> 00:18:25.400
and the top three ideas will get a B device and I think we also have one for crowd favorite.

s184
00:18:25.400 --> 00:18:30.720
So, with that going to jump into the workshop.

s185
00:18:30.720 --> 00:18:34.840
This is the link for the workshop.

s186
00:18:34.840 --> 00:18:41.880
You can type in it's pretty it's it's available on GitHub so you can type in perception agents workshop.

s187
00:18:41.880 --> 00:18:45.080
You will probably get the first GitHub link.

s188
00:18:45.080 --> 00:18:54.760
Um in the workshop we also need you to have use a Nova Act API key.

s189
00:18:54.760 --> 00:19:01.840
So, there is a piece where in the workshop we have obscured the first three letters.

s190
00:19:01.840 --> 00:19:04.280
So, that is on the screen here.

s191
00:19:04.280 --> 00:19:15.000
So, you can like copy paste these three letters ACE that stands for ace and you're good to go with the Novact API key.

s192
00:19:17.120 --> 00:19:24.200
So, let me just pause this.

s193
00:19:24.200 --> 00:19:30.040
So, in the workshop, you would the readme will basic point you to the workshop with MD file.

s194
00:19:30.040 --> 00:19:33.320
And here is the scenario that I was talking about.

s195
00:19:33.320 --> 00:19:34.560
I'm going to read it out from here.

s196
00:19:34.560 --> 00:19:40.160
You are a front-end developer preparing to ship a landing page for a podcast.

s197
00:19:40.160 --> 00:19:46.000
Yesterday you demoed the page to a few peers and they provided some feedback.

s198
00:19:46.000 --> 00:19:48.560
The hero title feels too big.

s199
00:19:48.560 --> 00:19:51.560
The accent color doesn't pop up the way you it should.

s200
00:19:51.560 --> 00:19:58.000
Some buttons are oversized and a few small things that you slightly just off.

s201
00:19:58.000 --> 00:20:00.840
So, you are on board with those feedback.

s202
00:20:00.840 --> 00:20:03.320
And today you come in fresh.

s203
00:20:03.320 --> 00:20:07.160
And I'm ready to incorporate those changes.

s204
00:20:07.160 --> 00:20:09.720
So, what do you have at your disposal?

s205
00:20:09.720 --> 00:20:18.000
The Novact Chrome extension, the UI verification scale, you have the Novact SDK, and the Novact MCP server.

s206
00:20:18.000 --> 00:20:31.400
These are the tools that you have at your disposal to basically implement the solution where you would essentially build build this entire loop by itself.

s207
00:20:31.400 --> 00:20:42.080
Like, you just annotate and automatically you would just hit a button in the Chrome extension and all the changes will be applied.

s208
00:20:42.080 --> 00:20:47.760
The UI verification will run and you have a report by the end of it.

s209
00:20:47.760 --> 00:20:52.760
So, this is the challenge and there are a few prerequisites.

s210
00:20:52.760 --> 00:21:01.120
So, um you need to have an macOS or a Linux tool because Novact MCP server runs on macOS and Linux.

s211
00:21:01.120 --> 00:21:03.200
You should have Node.js installed.

s212
00:21:03.200 --> 00:21:05.000
If not, you can install them.

s213
00:21:05.000 --> 00:21:15.120
uh NPM, Python, uh Chrome browser to basically test it out because it's a Chrome extension, and then um an AI coding agent.

s214
00:21:15.120 --> 00:21:17.880
So, if you have Claude, that works great.

s215
00:21:17.880 --> 00:21:19.440
We also have Kiro integration.

s216
00:21:19.440 --> 00:21:26.600
So, if you don't have any co- coding um agent, you can install Kiro and start playing around with it.

s217
00:21:26.600 --> 00:21:31.120
Um and lastly, we have the UV um that's UVX.

s218
00:21:31.120 --> 00:21:32.840
That's basically to install.

s219
00:21:32.840 --> 00:21:39.880
If you don't want to install UV or UVX, then please raise your hand and we can help you install using pip.

s220
00:21:39.880 --> 00:21:44.520
Um So, that's that's that.

s221
00:21:44.520 --> 00:21:50.840
And the workshop is laid out for I think the setup should be about 5 to 7 minutes.

s222
00:21:50.840 --> 00:21:54.880
Uh then you can like start playing around with those skills.

s223
00:21:54.880 --> 00:22:01.480
That gets a good understanding of how would you use the Chrome extension, how would you use the uh UI verification skill.

s224
00:22:01.480 --> 00:22:03.680
That's what has been explained here.

s225
00:22:03.680 --> 00:22:13.760
Uh to solve the challenge, you have um about 15 minutes in if you want to go that route, like I want to go step by step.

s226
00:22:13.760 --> 00:22:16.520
And I'm just going to go scroll down real quick.

s227
00:22:16.520 --> 00:22:16.960
Yeah.

s228
00:22:16.960 --> 00:22:18.000
So, choose your path.

s229
00:22:18.000 --> 00:22:21.680
I have this workshop designed in two levels.

s230
00:22:21.680 --> 00:22:24.960
Uh you you remember I said like I want you to have fun.

s231
00:22:24.960 --> 00:22:28.640
So, two levels, if you want to go for a challenge, go for the medium route.

s232
00:22:28.640 --> 00:22:34.920
If you feel that uh I'm just going to like uh follow the instructions and get the workshop done.

s233
00:22:34.920 --> 00:22:36.040
You'll feel good about it.

s234
00:22:36.040 --> 00:22:46.280
So, yeah, we have an easy mode where you'll have exact instructions what you what should be done, and you will have these two primitives wired up by the end of the hands-on.

s235
00:22:46.280 --> 00:22:53.680
Um so, 15 minutes we are expecting the easy path, um and then 30 to 35 minutes for the medium.

s236
00:22:53.680 --> 00:23:06.400
Once we have allocated the 15 minutes time, I will come up and show you where you would like essentially um add your code so that uh you can wire these two primitives up

s237
00:23:06.400 --> 00:23:11.200
and you can see the entire thing come into life in action.

s238
00:23:11.840 --> 00:23:14.080
And we also have a bonus challenge.

s239
00:23:14.080 --> 00:23:17.280
If you finish early, you can start with the bonus challenge.

s240
00:23:17.280 --> 00:23:26.440
Uh but before that we will have the B team Giovanni um in some time to like present.

s241
00:23:27.360 --> 00:23:28.240
After Yeah.

s242
00:23:28.240 --> 00:23:29.280
Yeah, sure thing.

s243
00:23:29.280 --> 00:23:34.040
So I saw that you had as I was installing these uh

s244
00:23:34.040 --> 00:23:35.200
Mhm.

s245
00:23:35.200 --> 00:23:39.200
What is it the Nova Act UI verification skills?

s246
00:23:39.200 --> 00:23:46.840
Um it it said that there were some defaults clients that are always supported and it included like Codex and Cloud Code.

s247
00:23:46.840 --> 00:23:51.920
But then below that there are these additional skills and one of them is also Cloud Code.

s248
00:23:51.920 --> 00:23:57.960
So I'm just curious like are there two different things going on there or are those the same?

s249
00:23:57.960 --> 00:24:00.760
The the fact that there are two Cloud Code

s250
00:24:00.760 --> 00:24:02.960
Yes, so there are two skills.

s251
00:24:02.960 --> 00:24:05.480
One is the Nova Act skill.

s252
00:24:05.480 --> 00:24:07.600
One is the Nova Act UI verification skills.

s253
00:24:07.600 --> 00:24:09.560
So that's why you must be seeing both of them.

s254
00:24:09.560 --> 00:24:11.560
Yeah.

s255
00:24:13.560 --> 00:24:21.640
So I'm going to give you the time here to I'm going to bring up the slide back.

s256
00:24:21.640 --> 00:24:23.920
I'm going to ask Emil to pop up here.

s257
00:24:23.920 --> 00:24:29.440
Slide back on the GitHub repo and so that you have the A API key handy if you have to change.

s258
00:24:29.440 --> 00:24:33.320
Feel free to like raise your hand if you have any questions.

s259
00:24:33.320 --> 00:24:37.640
We have our great awesome team supporting the workshop.

s260
00:24:37.640 --> 00:24:40.080
There you find them, see them waving.

s261
00:24:40.080 --> 00:24:47.280
So reach out to either of us and we will pop by you to help.

s262
00:24:47.280 --> 00:24:49.560
Have fun.

s263
00:24:49.840 --> 00:25:00.960
Okay, I'm just going to try to walk through the solution so that Okay.

s264
00:25:12.360 --> 00:25:15.040
So, this is the step-by-step instructions.

s265
00:25:15.040 --> 00:25:18.280
You can see that we have the Chrome extension.

s266
00:25:18.280 --> 00:25:22.920
We have your agent uh AI agent running.

s267
00:25:22.920 --> 00:25:33.040
Uh if you want your coding agent to like automatically apply the changes from the extension, you need a bridge because there's no bridge between them.

s268
00:25:33.040 --> 00:25:37.760
How would the extension like send the prompt to the coding uh coding agent?

s269
00:25:37.760 --> 00:25:50.960
So, for that, we have built uh something called agent bridge that will automatically uh take the annotations and up- give prompted to the coding agent.

s270
00:25:50.960 --> 00:25:56.680
That's where it will start applying those changes and then running the UI verification skill.

s271
00:25:56.680 --> 00:26:12.920
So, um essentially in the extension, we don't have the apply changes button in the in the primitive mode, but I for this workshop, I have built in we have built this um apply changes button

s272
00:26:12.920 --> 00:26:14.640
in the Chrome extension.

s273
00:26:14.640 --> 00:26:18.240
Um I'll walk you through that in a bit.

s274
00:26:18.240 --> 00:26:26.120
Uh so, for the Chrome for for this entire thing to work, you need to make changes in like two files.

s275
00:26:26.120 --> 00:26:42.920
The first file is um the manifest.json file in the Chrome extension, where you essentially add the this line for the host permissions to start accepting requests from the local host.

s276
00:26:42.920 --> 00:26:47.560
Uh and then you add this this piece of code.

s277
00:26:47.560 --> 00:26:52.240
It's a huge like piece of chunk of code here.

s278
00:26:52.240 --> 00:27:05.680
Um, and you can what this part does in the extension is creates and apply changes button, wires up with the extension, uh, make sure makes makes the UI verification

s279
00:27:05.680 --> 00:27:17.920
run uh, in using the Nova at SDK instead of the skill, uh, and starts generates a report that you can click on and then view the report.

s280
00:27:17.920 --> 00:27:35.200
Um, so this piece of code is goes in this file that is the, um, let's see, content.js in the extension and it needs to be added around like line 261.

s281
00:27:35.200 --> 00:27:48.160
Uh, between this append child export button just before and I think I have hash, uh, there's a comment there add code here in that file and that's where you would like paste this code.

s282
00:27:48.240 --> 00:27:59.640
So once you have that running now you would see this apply change button here in the extension.

s283
00:27:59.640 --> 00:28:07.520
Um, I think, uh, Emil was testing this so he changed this to like FIFA World Cup, why not?

s284
00:28:07.520 --> 00:28:18.440
Um, so I'm going to select one of these and I'm going to refresh the extension again.

s285
00:28:20.080 --> 00:28:22.520
Let's see.

s286
00:28:25.080 --> 00:28:28.080
Annotate current page.

s287
00:28:28.080 --> 00:28:32.800
Uh, I'm going to try to change the published date.

s288
00:28:34.080 --> 00:28:42.440
Change to today's June 20 9th.

s289
00:28:42.440 --> 00:28:43.840
Okay.

s290
00:28:43.840 --> 00:28:46.360
And save it.

s291
00:28:46.360 --> 00:28:50.720
And this is the extension.

s292
00:28:50.720 --> 00:28:52.600
Apply changes.

s293
00:28:52.600 --> 00:28:56.560
So, in the background, it's going to open up your coding agent.

s294
00:28:56.560 --> 00:29:00.400
It's been tested on Cloud Code and Kiro as I mentioned.

s295
00:29:00.400 --> 00:29:14.200
So, if you're trying Cursor or Codex, the agent bridge is something that you would have to edit so that it can accept these apply changes button like requests.

s296
00:29:14.200 --> 00:29:24.240
Um So, in a bit it'll start applying the changes and you will see certain logs that pop up as well in the repository.

s297
00:29:24.240 --> 00:29:36.800
Um that is the key the CLI's logs as well as when the verification runs Nova Act has these logs uh that gives you exactly what the AI model was thinking

s298
00:29:36.800 --> 00:29:40.920
and what part of the screen it was trying to make a decision on.

s299
00:29:40.920 --> 00:29:44.480
And you can see that the button changed to run verification now.

s300
00:29:44.480 --> 00:29:47.720
So, it's running the verification step.

s301
00:29:48.400 --> 00:29:51.680
And in a bit it'll also generate a report that you can see.

s302
00:29:51.680 --> 00:30:04.440
So, this is where you basically apply the changes and go grab a lunch or like break and then come back and just see uh the verification report once you're back.

s303
00:30:05.240 --> 00:30:13.280
Going to give it a minute because it's going to go through all the design flows, all the user flows, and then generate a report.

s304
00:30:13.280 --> 00:30:16.480
So, it takes a bit of a time.

s305
00:30:19.440 --> 00:30:22.880
Let's see if I can pull up the repo.

s306
00:30:24.520 --> 00:30:25.640
Oh.

s307
00:30:25.640 --> 00:30:27.640
Shoot.

s308
00:31:35.360 --> 00:31:52.440
I think the Wi-Fi is super slow, so But, I I think maybe in the meantime I can have Giovanni come over on the stage and show you a cool way to annotate your webs web pages

s309
00:31:52.440 --> 00:31:59.840
instead of like just going and typing stuff on the web on the web page that you're working on.

s310
00:31:59.840 --> 00:32:00.520
Hello.

s311
00:32:00.520 --> 00:32:01.400
Oh, the prime.

s312
00:32:01.400 --> 00:32:02.640
Okay.

s313
00:32:02.640 --> 00:32:04.200
Can we have

s314
00:32:04.200 --> 00:32:05.240
Uh I'm done.

s315
00:32:05.240 --> 00:32:07.080
Can I see the slides?

s316
00:32:07.080 --> 00:32:09.560
The slides, yes.

s317
00:32:11.360 --> 00:32:13.360
Hello.

s318
00:32:14.840 --> 00:32:26.840
Hi everyone, I'm Giovanni and I was thinking to imagine you are in to do your design meeting just talking about the design modification you want to do to that website.

s319
00:32:26.840 --> 00:32:32.400
And then automatically your perception agent agent is able to modify the website.

s320
00:32:32.400 --> 00:32:35.440
This is something that you can do right now with the workshop.

s321
00:32:35.440 --> 00:32:37.600
And this why, how?

s322
00:32:37.600 --> 00:32:40.520
Using the B wearable devices.

s323
00:32:40.520 --> 00:32:43.640
So, B is a device we have in Amazon.

s324
00:32:43.640 --> 00:32:47.920
And you can buy it like in the b.computer website.

s325
00:32:47.920 --> 00:32:50.600
It's a wearable this is your AI assistant.

s326
00:32:50.600 --> 00:32:53.640
So, it just have all it has a microphone.

s327
00:32:53.640 --> 00:33:00.040
So, you can activate this the button and then it's going to listen for your conversation.

s328
00:33:00.040 --> 00:33:11.800
No worries, there is a l l i d that's will be on if it's listening to your conversation and then all your conversation on your private cloud.

s329
00:33:11.800 --> 00:33:21.000
After the AI agent that is running B is able to gives you suggestion about whatever conversation you had.

s330
00:33:21.000 --> 00:33:28.960
So, what I did is like integrating with this this agent the B agent with the perception agent that we had for the workshop.

s331
00:33:28.960 --> 00:33:36.080
So, getting the input listening from a AI conversation from a design conversation.

s332
00:33:36.080 --> 00:33:38.760
Uh, I don't see the video.

s333
00:33:38.760 --> 00:33:40.000
Yeah.

s334
00:33:40.000 --> 00:33:40.720
Ah, this one.

s335
00:33:40.720 --> 00:33:41.480
Yeah.

s336
00:33:41.480 --> 00:33:43.720
Thank you.

s337
00:33:43.960 --> 00:33:46.600
You can achieve this with three simple step.

s338
00:33:46.600 --> 00:33:54.080
So, first of all, yes, you have to have a B. And I want with me that you can use here and now or we can have a test a test account

s339
00:33:54.080 --> 00:33:58.680
to download that you can use to get the conversation as you want the conversation.

s340
00:33:58.680 --> 00:34:06.080
Then you need to install the B C L I. It's a C L I that will enable that is going to allow you to get the data for your conversation

s341
00:34:06.080 --> 00:34:11.440
and it's in the workshop the instruction to download the B C L I by just a npm command.

s342
00:34:11.440 --> 00:34:23.480
Then you start the server of the workshop and then you start the bridge listening from B conversation and applying that to the perception agent.

s343
00:34:23.480 --> 00:34:25.800
Then you just enable your B device.

s344
00:34:25.800 --> 00:34:27.240
The LED is on.

s345
00:34:27.240 --> 00:34:30.919
You say something like, "Oh, I want to change the title color to red.

s346
00:34:30.919 --> 00:34:35.240
I want to change the emoji instead of our brain emoji, I want a robot emoji.

s347
00:34:35.240 --> 00:34:42.919
And on the right side, you will see your conversation, your transcript of the conversation, and even the most important points.

s348
00:34:42.919 --> 00:34:50.639
Here you can see the response API, the the API response coming from the B. You can open the panel and check.

s349
00:34:50.639 --> 00:34:53.120
And then you can just select apply.

s350
00:34:53.120 --> 00:34:57.600
And here is going to be the normal flow of the perception agent workshop.

s351
00:34:57.600 --> 00:35:03.920
So, it's going to change the color, the color that you just said, it's going to change the emoji.

s352
00:35:03.920 --> 00:35:16.640
And after the modification of the website are applied, it's going to run even for you automatically the verification phase of no act.

s353
00:35:16.640 --> 00:35:23.400
So, we have a test account that you can use with a preloaded conversation, preloaded design conversation.

s354
00:35:23.400 --> 00:35:27.320
It is like just talking about, "Oh, we change the color, we change the emoji."

s355
00:35:27.320 --> 00:35:29.120
Something like that.

s356
00:35:29.120 --> 00:35:38.480
And you can after you install the CLI and you log in with the B account with a test device with a test account, you just click on the bottom

s357
00:35:38.480 --> 00:35:41.760
button, the the button at the bottom.

s358
00:35:41.760 --> 00:35:44.440
Uh this is the verification page you can check.

s359
00:35:44.440 --> 00:35:51.080
But otherwise, if you want to test the B device, we can like you do we can actually just use mine and I will show you around if you want it.

s360
00:35:51.080 --> 00:36:01.400
Like we can actually see like have a conversation about the design and you can see that it's actually applying this after like two or three minutes of elaborating the conversation.

s361
00:36:01.400 --> 00:36:04.520
The Wi-Fi is a little bit slow, so that's why I didn't do the live demo.

s362
00:36:04.520 --> 00:36:09.360
Uh and yeah, and then in the end there is there is the verification report that you can check.

s363
00:36:09.360 --> 00:36:13.880
The B devices are also our prizes for the for the challenge.

s364
00:36:13.880 --> 00:36:16.920
So, yeah, you're going to have one if you win the challenge.

s365
00:36:16.920 --> 00:36:21.920
And we also are at the booth today and all the week.

s366
00:36:22.080 --> 00:36:26.760
So, yeah, let me know if you want to integrate B and I'm here for you.

s367
00:36:26.760 --> 00:36:40.480
So, B device, if you want to come up on the stage, we have 10 minutes for you to like explore these two primitives, think of an idea that you want to probably use it.

s368
00:36:40.480 --> 00:36:43.960
And if you want to come up, you'll get one of these devices.

s369
00:36:43.960 --> 00:36:46.000
Um so, yeah.

s370
00:36:46.000 --> 00:36:52.200
The if just get creative, I guess.

s371
00:36:52.240 --> 00:36:59.160
Hey everyone, I have Jed with me who built something really cool and he would like to show it to everyone.

s372
00:36:59.160 --> 00:37:01.840
Jed, come up on the stage.

s373
00:37:02.720 --> 00:37:04.320
Yeah.

s374
00:37:04.320 --> 00:37:05.000
Jed.

s375
00:37:05.000 --> 00:37:05.800
Oh.

s376
00:37:05.800 --> 00:37:06.240
What am I doing?

s377
00:37:06.240 --> 00:37:07.000
Where am I going?

s378
00:37:07.000 --> 00:37:08.400
Oh, I called Jed, but yeah.

s379
00:37:08.400 --> 00:37:09.960
Oh, I I My name is Jeff, so I thought

s380
00:37:09.960 --> 00:37:12.000
You are Sorry.

s381
00:37:12.000 --> 00:37:15.880
I'll have Jeffrey next to show something cool as well.

s382
00:37:15.880 --> 00:37:17.240
Oh, okay.

s383
00:37:17.240 --> 00:37:32.600
All right, this this was just goofy, but it's actually an an an interesting little funnel here is is with this animators going through and Oh, you can't see it.

s384
00:37:33.600 --> 00:37:36.440
Change it to a duck podcast.

s385
00:37:40.800 --> 00:37:43.080
There we go.

s386
00:37:43.960 --> 00:37:48.000
So, just turn asked it to turn it retro NES.

s387
00:37:48.000 --> 00:37:55.120
Uh do some duck jokes and went back and did the verification funnel.

s388
00:37:55.120 --> 00:38:08.600
Um it's probably as what I was telling them is probably really interesting to me here is to watch this this great AI said I made it a duck you know, I I made a duck icon.

s389
00:38:08.600 --> 00:38:22.120
But uh often as everyone here likely knows is you don't actually see that happen, but then it it went through and did the the entire verification funnel, came back with the reports

s390
00:38:22.120 --> 00:38:23.480
saying, yep, it's done.

s391
00:38:23.480 --> 00:38:24.160
I did it.

s392
00:38:24.160 --> 00:38:25.120
I actually did it.

s393
00:38:25.120 --> 00:38:27.920
So, that was cool.

s394
00:38:29.720 --> 00:38:31.920
Oh, yeah.

s395
00:38:33.920 --> 00:38:38.720
I had it I had it just do asked it to do some jokes.

s396
00:38:39.840 --> 00:38:42.600
Which is pretty bad at.

s397
00:38:45.520 --> 00:38:47.720
There you go.

s398
00:38:52.960 --> 00:38:57.160
Oh, because you came up with a cool demo.

s399
00:38:57.160 --> 00:38:59.560
Do you have to step?

s400
00:39:01.280 --> 00:39:03.400
As promised, you win this.

s401
00:39:03.400 --> 00:39:04.240
Oh, thank you.

s402
00:39:04.240 --> 00:39:06.400
P device.

s403
00:39:10.240 --> 00:39:17.080
UP NEXT, WE HAVE Jeff Lee who wants to come and present what he did.

s404
00:40:00.880 --> 00:40:10.680
Yeah, so I'm still building a test harness to actually be able to show this cuz the default B account has a uh sample conversation that isn't relevant to the things.

s405
00:40:10.680 --> 00:40:14.960
So, this is just this is just something trying to uh bring the B to life.

s406
00:40:14.960 --> 00:40:28.360
So, the idea was to have access the BCLI um and collect the summary of a of a conversation which was a discussion of of a backlog features that would be added to a

s407
00:40:28.360 --> 00:40:30.280
uh repo that you're working on, right?

s408
00:40:30.280 --> 00:40:41.360
So, you're working on a specific uh GitHub repository that's that's that that's cloned and then um uh what's going to happen is your conversation is going to be used in context with

s409
00:40:41.360 --> 00:40:43.280
uh all of the code that's that's there.

s410
00:40:43.280 --> 00:40:50.720
It's going to build a backlog based on the the summary and it's going to use the Atlassian MCP to put that into Jira.

s411
00:40:50.720 --> 00:40:54.320
So, that's the uh that's the uh idea here.

s412
00:40:54.320 --> 00:41:02.640
And so, the key thing that I'm just building right now in the background is uh the uh the test harness to insert a new conversation that's relevant to the the repo.

s413
00:41:02.640 --> 00:41:11.160
So, that's Pretty simple and uh yeah.

s414
00:41:11.160 --> 00:41:12.280
Anyway.

s415
00:41:12.280 --> 00:41:13.000
That's awesome.

s416
00:41:13.000 --> 00:41:13.960
Thank you so much.

s417
00:41:13.960 --> 00:41:15.960
Thanks.

s418
00:41:17.160 --> 00:41:17.480
You're welcome.

s419
00:41:17.480 --> 00:41:19.280
So, here you win your new device.

s420
00:41:19.280 --> 00:41:22.040
I know you wanted to go and get it, but you won it.

s421
00:41:22.040 --> 00:41:24.320
Thank you very much.

s422
00:41:29.960 --> 00:41:34.760
So, You have a question?

s423
00:41:37.480 --> 00:41:41.240
Ryan, who's going to present next?

s424
00:41:43.640 --> 00:41:45.040
Hello, everyone.

s425
00:41:45.040 --> 00:41:47.480
Excited to be here.

s426
00:41:53.840 --> 00:41:54.720
Let me see.

s427
00:41:54.720 --> 00:41:56.280
Start mirroring.

s428
00:41:56.280 --> 00:42:02.200
So, I was I was late to this session uh because the other session ended at 3:00.

s429
00:42:02.200 --> 00:42:07.600
So, um I came up to uh Giovanni.

s430
00:42:07.600 --> 00:42:11.360
Let me shift the key right here.

s431
00:42:12.360 --> 00:42:13.640
Oh, perfect.

s432
00:42:13.640 --> 00:42:14.840
That's fine.

s433
00:42:14.840 --> 00:42:16.080
So, I gave it to Giovanni.

s434
00:42:16.080 --> 00:42:18.800
Um currently wearing a bee hat.

s435
00:42:18.800 --> 00:42:22.240
I currently attend uh Wharton uh business school.

s436
00:42:22.240 --> 00:42:24.120
My cluster's mascot is a bee.

s437
00:42:24.120 --> 00:42:28.240
I am one of three people who care that the mascot to be.

s438
00:42:28.240 --> 00:42:34.520
Um so, you can imagine my excitement when I came to this conference and saw we were presenting about uh bee device.

s439
00:42:34.520 --> 00:42:40.520
So, naturally, being late, being a Wharton grad, I went and asked if I could have a free bee device.

s440
00:42:40.520 --> 00:42:43.520
Uh Giovanni said, "You have an idea to share?"

s441
00:42:43.520 --> 00:42:46.920
So, I I currently working at an AI upskilling startup.

s442
00:42:46.920 --> 00:42:49.800
Um I'm not a huge fan of Linear.

s443
00:42:49.800 --> 00:42:52.440
Apologies to anyone who works at Linear.

s444
00:42:52.440 --> 00:43:03.640
I just needed a better way of understanding what should I be doing anytime during the day, going from capturing an idea, discussing with my co-founder, and then updating a project repository.

s445
00:43:03.640 --> 00:43:06.440
So, this is a natural use case for the bee device.

s446
00:43:06.440 --> 00:43:09.360
So, we have a internal tool called Sidekick.

s447
00:43:09.360 --> 00:43:13.440
Emphasis on an internal because there is no design.

s448
00:43:13.440 --> 00:43:21.120
Um in my co-founder's defense, he did tell me to uh tell Claude to make it look pretty, no mistakes, and this is the best it did.

s449
00:43:21.120 --> 00:43:29.040
So, um but how it works is we have a a joint inbox.

s450
00:43:29.040 --> 00:43:31.760
Um you can ignore the Go Touch Crash SF.

s451
00:43:31.760 --> 00:43:35.320
I just see all these billboards and um gets me concerned.

s452
00:43:35.320 --> 00:43:39.840
Uh that we send emails jointly to uh capture@sidekick.ausin.org.

s453
00:43:41.000 --> 00:43:46.560
And these are all the different inputs we have um thoughts throughout the week, newsletters.

s454
00:43:46.560 --> 00:43:49.720
I don't know if anyone here is subscribed to an AI newsletter.

s455
00:43:49.720 --> 00:43:53.000
Um see something interesting, we send it to inbox.

s456
00:43:53.000 --> 00:44:03.080
And we are big believers in getting things done, which was a uh mode for knowledge workers of organizing a projects task developed by David Allen in 2010.

s457
00:44:03.080 --> 00:44:07.400
And oh, if the Wi-Fi works, you could be able to see it.

s458
00:44:07.400 --> 00:44:12.520
Um Sorry, that did just work off the screen.

s459
00:44:12.920 --> 00:44:13.840
One second.

s460
00:44:13.840 --> 00:44:27.760
So, the getting the getting things done tree shows all the projects, and under each project it has uh the strategy, has what the what completion looks like, tasks, next steps, references.

s461
00:44:27.760 --> 00:44:34.520
And we've built an MCP so we can sit and talk to Claude and strategize on, "Hey, this is what we're thinking on this project.

s462
00:44:34.520 --> 00:44:37.040
What do you think we should be doing?"

s463
00:44:37.040 --> 00:44:43.640
All to solve the need of when you're working after you have a meeting of like, "What should I be doing right now to tie to my ultimate strategy?"

s464
00:44:43.640 --> 00:44:58.520
So, how we imagine this to work with the uh B uh device, as Giovanni uh informed me there's an MCP, would be capture different meetings or different ideas, have it sent to our capture inbox, and then use that to update our strategy.

s465
00:44:58.520 --> 00:45:02.760
Um I wish this was was loading because, let me see.

s466
00:45:02.760 --> 00:45:04.800
Oh, that's why.

s467
00:45:04.800 --> 00:45:06.400
my phones.

s468
00:45:06.400 --> 00:45:07.920
See that worked.

s469
00:45:07.920 --> 00:45:11.720
Um Yes.

s470
00:45:11.720 --> 00:45:15.120
You know, I'm now 0 for 11 for live demos.

s471
00:45:15.120 --> 00:45:19.080
Um but that is that is my concept of the B device.

s472
00:45:19.080 --> 00:45:24.400
I'm looking forward to using it and sharing on LinkedIn how much I love the bees.

s473
00:45:24.400 --> 00:45:25.920
Uh go bees.

s474
00:45:25.920 --> 00:45:28.120
Buzz buzz.

s475
00:45:28.480 --> 00:45:30.680
Thank you.

s476
00:45:34.040 --> 00:45:41.040
How would you uh go and integrate the perception agents into your uh

s477
00:45:41.040 --> 00:45:42.760
Oh, I actually I have an answer for that, too.

s478
00:45:42.760 --> 00:45:43.720
Okay.

s479
00:45:43.720 --> 00:45:55.680
So, as I've actually emailed myself inside kick to discuss later with my co-founder, is uh we developed a lot of product offerings but such as sidekick, um different tools for learning.

s480
00:45:55.680 --> 00:46:01.680
Um the whole mission of making sure everyone knows how to adjust to the upcoming workforce displacement.

s481
00:46:01.680 --> 00:46:09.120
But the bottleneck right now is product design of building something that users can understand exactly what to do.

s482
00:46:09.120 --> 00:46:22.120
So, we would use the annotation feature of the perception agent to then um send that description either through sidekick through our own like agent we have running to to make improvements, but then to close that loop of development.

s483
00:46:22.120 --> 00:46:22.480
Oh.

s484
00:46:22.480 --> 00:46:23.600
So.

s485
00:46:23.600 --> 00:46:28.840
Yeah, so if AGI would love to sponsor Ozzie and I will give you my card and we can talk later.

s486
00:46:28.840 --> 00:46:29.480
So.

s487
00:46:29.480 --> 00:46:30.280
Thank you.

s488
00:46:30.280 --> 00:46:30.800
Awesome.

s489
00:46:30.800 --> 00:46:35.920
Thank you so much for coming up and demoing and here as promised a B device.

s490
00:46:35.920 --> 00:46:36.160
Thank you.

s491
00:46:36.160 --> 00:46:38.560
Appreciate it.

s492
00:46:39.960 --> 00:46:51.480
So, as we are wrapping this up, um just Emil, I'm just going to We are going to put up a survey on the board on the screen.

s493
00:46:51.480 --> 00:46:58.680
Uh it's something that we Yes.

s494
00:46:59.560 --> 00:47:01.840
You want to Oh, okay.

s495
00:47:01.840 --> 00:47:08.240
Okay, there is uh Jonathan who wants to present as well.

s496
00:47:11.440 --> 00:47:16.280
Uh so, mine's more of like a concept that I didn't have time to build everything.

s497
00:47:16.280 --> 00:47:20.000
Um but I'll just kind of walk through it.

s498
00:47:30.240 --> 00:47:30.520
All right.

s499
00:47:30.520 --> 00:47:39.160
So, I was kind of looking at the uh the proxy for the B and trying to figure out how are we getting the information from the CLI.

s500
00:47:39.160 --> 00:47:41.560
Oh, it's not.

s501
00:47:41.560 --> 00:47:42.200
It's not sharing.

s502
00:47:42.200 --> 00:47:42.800
Oh, there we go.

s503
00:47:42.800 --> 00:47:43.600
All right.

s504
00:47:43.600 --> 00:47:56.360
So, I was trying to figure out uh how the proxy actually works and how we're getting information from the B and uh besides just doing transcripts uh transcription could you pull different data from different sources?

s505
00:47:56.360 --> 00:47:59.560
Could you pull up from an IoT device for whatever reason?

s506
00:47:59.560 --> 00:48:07.400
Um And what I started to come up with is uh at the company I'm at, which is a finance company, we have a lot of non-technical people.

s507
00:48:07.400 --> 00:48:19.640
And not only do we have to get our uh criteria in, we also have like a lot of UI UX um that we have to try and capture, and we don't always get it in like the best

s508
00:48:19.640 --> 00:48:23.120
details uh for acceptance criteria.

s509
00:48:23.120 --> 00:48:35.720
So, I've been pushing the team to use uh Jam MCP, which basically allows you to do a recording, captures all the console logs, captures the network requests, and then uh feed that information.

s510
00:48:35.720 --> 00:48:41.160
Um we also have I built an AR AI harness so we can pipe this information to it.

s511
00:48:41.160 --> 00:48:53.560
So, really the architecture kind of looks like this, where very similar to like the um the CLI, you could actually take the Jam MCP, pipe that information into the proxy,

s512
00:48:53.560 --> 00:49:00.240
have the LLM go out to you um Nova, and then actually take the UI uh implementation.

s513
00:49:00.240 --> 00:49:11.960
So, if you're saying things like, "Oh, I wish this uh would say, you know, uh this investor schedule needs to go like this um or this score should be

s514
00:49:11.960 --> 00:49:16.680
uh a research level score and be in a different location."

s515
00:49:16.680 --> 00:49:25.320
Uh we'd be able to actually pull that information from the MCP, send it over to Nova, and actually make the changes on the AI harness that is running uh locally.

s516
00:49:25.320 --> 00:49:30.800
So, you could do the proxy the same way, um but through a different device.

s517
00:49:30.840 --> 00:49:32.520
Um did some initial tests.

s518
00:49:32.520 --> 00:49:41.240
It looks like it works even through just some mocking, uh but wasn't able with the Wi-Fi to get uh the fig or the Jam MCP up and running.

s519
00:49:41.240 --> 00:49:43.920
But, that'll be my test when I get home.

s520
00:49:43.920 --> 00:49:45.160
So.

s521
00:49:45.160 --> 00:49:47.160
Thanks.

s522
00:49:54.520 --> 00:49:58.440
Thank you so much and here you go.

s523
00:49:58.920 --> 00:50:06.520
So, right now we have 15 more minutes that will close the workshop.

s524
00:50:06.520 --> 00:50:12.360
Um we will put up a QR code on screen.

s525
00:50:12.360 --> 00:50:25.280
If you feel that you have a great idea and you don't want to come up and talk to us about it you can submit these ideas here and we'll do a

s526
00:50:26.560 --> 00:50:33.520
we will sub we will vote on those ideas and the top three ideas will get a TB device.

s527
00:50:33.520 --> 00:50:43.000
So, if you have an idea please put it in those the poll.

s528
00:50:43.480 --> 00:50:46.680
Just give me a second.

s529
00:50:47.680 --> 00:50:48.240
There you go.

s530
00:50:48.240 --> 00:50:49.080
You guys see that?

s531
00:50:49.080 --> 00:50:51.080
Yeah.

s532
00:50:52.960 --> 00:50:59.720
So, we want them to scan the QR code and enter an innovative idea using what we've learned.

s533
00:50:59.720 --> 00:51:07.120
What would you use this perception, be all the things that you saw today, all the things you played around with today?

s534
00:51:07.120 --> 00:51:17.120
What do you think would be a good use case for you that will help you make your life easier on a everyday basis?

s535
00:51:19.160 --> 00:51:22.240
We saw a bunch of ideas.

s536
00:51:22.240 --> 00:51:32.320
Now, and if you think you had one you're feeling a little shy there's a chance to win another device.

s537
00:51:33.560 --> 00:51:34.000
So, what do you think?

s538
00:51:34.000 --> 00:51:35.560
5 minutes and then we'll

s539
00:51:35.560 --> 00:51:39.920
5 minutes and then we will Yep.

s540
00:51:44.600 --> 00:51:47.040
Let me fix that.

s541
00:51:49.240 --> 00:51:53.360
Okay, I stopped the poll now.

s542
00:51:53.360 --> 00:51:56.360
So, all of them great ideas.

s543
00:51:56.360 --> 00:52:01.800
I was reading through them and I was discussing with the mail there like so hard to pick.

s544
00:52:01.800 --> 00:52:04.080
Is Isn't it a mail?

s545
00:52:04.080 --> 00:52:16.640
So, we have the one with pixel judge.

s546
00:52:17.040 --> 00:52:20.720
Can someone raise their hand who submitted that idea?

s547
00:52:22.760 --> 00:52:24.920
No one?

s548
00:52:25.200 --> 00:52:27.560
Pixel judge?

s549
00:52:27.680 --> 00:52:31.320
Uh What's that?

s550
00:52:33.080 --> 00:52:34.400
Okay.

s551
00:52:34.400 --> 00:52:41.440
Um the next one was the agentic end-to-end testing for a product.

s552
00:52:41.440 --> 00:52:42.520
Okay.

s553
00:52:42.520 --> 00:52:44.960
Come on up here.

s554
00:52:48.760 --> 00:52:54.560
Do you want to have Yeah.

s555
00:52:54.560 --> 00:52:56.560
Awesome.

s556
00:52:56.720 --> 00:52:59.040
Thank you.

s557
00:53:00.920 --> 00:53:09.680
The reframe a life thinking canvas for brainstorming and planning.

s558
00:53:09.880 --> 00:53:10.600
That's you?

s559
00:53:10.600 --> 00:53:13.040
Come on up here.

s560
00:53:19.760 --> 00:53:28.520
So, we had decided we'll give away three and one was not claimed.

s561
00:53:28.520 --> 00:53:33.560
So, we have we did actually circle down to four.

s562
00:53:33.960 --> 00:53:42.840
The last one is build a canvas app where human perception agent work on a UI design together in real time.

s563
00:53:42.840 --> 00:53:45.280
Go on, up here.

s564
00:53:48.440 --> 00:53:51.120
Thank one.

s565
00:53:52.880 --> 00:53:58.040
A huge round of applause for all the winners here.

s566
00:54:12.280 --> 00:54:20.080
Okay, so I just want to wrap the session up with uh something like a key takeaway.

s567
00:54:20.080 --> 00:54:30.840
The first the most important thing of this workshop, you can all imagine, is the way human and agent collaboration is happening is changing.

s568
00:54:30.840 --> 00:54:45.040
So, we need tools that will make the visual piece of how the um human and agents are collaborating, and this is what a step next to building perception perception agent is.

s569
00:54:45.040 --> 00:54:51.240
So, you can get started with Nova Act if you haven't already, you know, looked at the GitHub repo.

s570
00:54:51.240 --> 00:54:54.320
It's available, you can scan this QR code.

s571
00:54:54.320 --> 00:55:06.640
It's an SDK, and also we have a blog post by our team that you can look it up explaining how what we are thinking about perception agent at large.

s572
00:55:06.680 --> 00:55:13.320
So, with that, thank you everyone, and we also have a feedback um QR code up.

s573
00:55:13.320 --> 00:55:22.120
If you think you like the session, you like Emil, uh you liked me, you liked Giovanni here who presented, please leave us a feedback.

s574
00:55:22.120 --> 00:55:24.200
We appreciate all your presence here.

s575
00:55:24.200 --> 00:55:26.680
Thank you so much.

s576
00:55:42.640 --> 00:55:44.640
Mhm.
