How AI Changed Software Development Forever — Building the Agentic Future with Andre Baltieri [MVP]

1
00:00:00,000 --> 00:00:05,600
Welcome to the MC65 with our podcast.

2
00:00:05,600 --> 00:00:10,480
Artifactil intelligence is changing software development and speed.

3
00:00:10,480 --> 00:00:16,200
We have rarely seen before for decades the developers wrote applications by defining exactly

4
00:00:16,200 --> 00:00:22,920
what software should do inputs, business rules, updates, database, interfaces, outputs,

5
00:00:22,920 --> 00:00:28,640
then generated on iLive and suddenly the developers could generate code, explain antimalosystems,

6
00:00:28,640 --> 00:00:35,120
create test, debug applications, interact with development environments using natural language.

7
00:00:35,120 --> 00:00:38,640
But that may only be the beginning.

8
00:00:38,640 --> 00:00:46,640
We are now moving forward an AI agent software that can reason about goal using tools and

9
00:00:46,640 --> 00:00:52,040
AI's main and daily context, interact with other agents and put it to execute complex

10
00:00:52,040 --> 00:00:57,280
workflows with much less direct human instructions.

11
00:00:57,280 --> 00:01:00,560
So what does this mean for the developers?

12
00:01:00,560 --> 00:01:06,880
Do they simply make us faster or do it fundamentally change what software, deployment,

13
00:01:06,880 --> 00:01:14,280
actually is and where do technology such as.net, Microsoft, agent, framework, AI models, tools

14
00:01:14,280 --> 00:01:19,840
and so on, and the architect just fit into the new world.

15
00:01:19,840 --> 00:01:25,240
Today I'm joining the Andre Balthary, Microsoft MVP software development specialist with

16
00:01:25,240 --> 00:01:28,560
more than two decades of experience in development.

17
00:01:28,560 --> 00:01:34,320
Andre has also worked with teams across Brazil, the United States, India, England and the Netherlands.

18
00:01:34,320 --> 00:01:40,840
And he also received the Microsoft MVP of our firm, Preston, 40 times for his contribution

19
00:01:40,840 --> 00:01:47,680
to the technical community, his areas of expertise, including artifact intelligence, the Microsoft

20
00:01:47,680 --> 00:01:50,760
agent, work.net and software development.

21
00:01:50,760 --> 00:01:54,840
And we welcome to the Microsoft C6-5 of M podcast.

22
00:01:54,840 --> 00:01:55,840
Thank you, Miracle.

23
00:01:55,840 --> 00:01:57,760
It's pleasure to be here today.

24
00:01:57,760 --> 00:01:59,760
I thank you.

25
00:01:59,760 --> 00:02:00,760
It's awesome.

26
00:02:00,760 --> 00:02:01,760
So, 40 times.

27
00:02:01,760 --> 00:02:02,760
Yeah.

28
00:02:02,760 --> 00:02:03,760
40 times.

29
00:02:03,760 --> 00:02:04,760
That's a long time.

30
00:02:04,760 --> 00:02:05,760
Yeah.

31
00:02:05,760 --> 00:02:13,280
Before we talk about AI, take us back to the 2003.

32
00:02:13,280 --> 00:02:17,120
What did the software development look like when you started?

33
00:02:17,120 --> 00:02:18,800
Yeah, well, that's nice.

34
00:02:18,800 --> 00:02:27,840
It's a nice question because I started with dot-ads in 2003 since the beginning of dot-ads

35
00:02:27,840 --> 00:02:29,320
basically.

36
00:02:29,320 --> 00:02:31,400
And things were completely different.

37
00:02:31,400 --> 00:02:38,800
So, at that time, we did not have so many information about software development.

38
00:02:38,800 --> 00:02:44,480
So, the internet itself was being created.

39
00:02:44,480 --> 00:02:50,800
It was very, very hard to learn something new at that time.

40
00:02:50,800 --> 00:02:59,120
I remember that I had to purchase some official Microsoft training, travel to another city

41
00:02:59,120 --> 00:03:04,680
at the weekends to do some training because it was crazy.

42
00:03:04,680 --> 00:03:10,040
You basically did not have access to any kind of information.

43
00:03:10,040 --> 00:03:25,800
And then, not sure if you are aware of it, but Microsoft had a very popular training

44
00:03:25,800 --> 00:03:30,040
center called MSDN, Microsoft Developer Network.

45
00:03:30,040 --> 00:03:31,040
Okay.

46
00:03:31,040 --> 00:03:33,720
It prevails to Microsoft to learn.

47
00:03:33,720 --> 00:03:39,960
It was the big place to go when you want to learn something from Microsoft.

48
00:03:39,960 --> 00:03:41,960
That was awesome.

49
00:03:41,960 --> 00:03:44,440
So, it all started there.

50
00:03:44,440 --> 00:03:51,760
So, since then, I was working with C#.net and it's very cool.

51
00:03:51,760 --> 00:03:53,240
It's my passion.

52
00:03:53,240 --> 00:03:54,240
Yeah.

53
00:03:54,240 --> 00:04:00,560
You have gone from desktop applications to the web, mobile, cloud, containers, Microsoft,

54
00:04:00,560 --> 00:04:08,600
serverless, and now AI do is that current AI transition feel different from those previous

55
00:04:08,600 --> 00:04:09,600
tests.

56
00:04:09,600 --> 00:04:10,600
Yeah.

57
00:04:10,600 --> 00:04:21,200
It's, it's sound like a new era, you know, because I, when I started, I was moving from desktop

58
00:04:21,200 --> 00:04:22,200
to web.

59
00:04:22,200 --> 00:04:29,040
This is the first era, they say, from software development.

60
00:04:29,040 --> 00:04:30,040
Everything was new.

61
00:04:30,040 --> 00:04:35,240
We were talking about Linux servers along.

62
00:04:35,240 --> 00:04:38,960
We, we do not ace.

63
00:04:38,960 --> 00:04:44,480
Then we have another big movement to mobile development.

64
00:04:44,480 --> 00:04:51,240
So, this was the new, nothing to create apps for iPhone and right.

65
00:04:51,240 --> 00:04:53,440
Then we have the cloud movement.

66
00:04:53,440 --> 00:05:04,440
When we move it to, to cloud, but nothing, so meaningful like move to AI, this, this

67
00:05:04,440 --> 00:05:13,680
era that we are passing today, because this is, for me, is the biggest change in software

68
00:05:13,680 --> 00:05:18,000
since the beginning, at least for my career.

69
00:05:18,000 --> 00:05:25,920
It's like, it's, it's like a new language, a new way to interact with, with software.

70
00:05:25,920 --> 00:05:37,160
So we basically move it from assembly or a very low language to a brand new language.

71
00:05:37,160 --> 00:05:43,640
Yeah, that's, that's, that's an amazing shift.

72
00:05:43,640 --> 00:05:50,520
Where, where we are today are AI coding tools, most of these sophisticated auto complete

73
00:05:50,520 --> 00:05:53,640
or are we already beyond set.

74
00:05:53,640 --> 00:06:02,720
Yeah, we're, I think we, we're in, in the middle, because I, I can see that we, we have

75
00:06:02,720 --> 00:06:06,560
much more to, to come yet to, to AI.

76
00:06:06,560 --> 00:06:18,240
But, AI is able to, to create a lot of things for us today, from the simple components to,

77
00:06:18,240 --> 00:06:25,440
high-calplex systems. So, they're evolving so fast because since we, we have a model that

78
00:06:25,440 --> 00:06:32,320
can create other models, so AI can evolve very fast.

79
00:06:32,320 --> 00:06:39,160
And, but, but this doesn't mean that we, we don't need programmers anymore.

80
00:06:39,160 --> 00:06:48,200
I think there's a bit of misconception out there, because AI can generate code for us.

81
00:06:48,200 --> 00:06:56,200
But, we need to ask the, the code generation in a certain form that it needs to, to understand,

82
00:06:56,200 --> 00:07:02,480
we need to understand the business concept, business rules before, uh, talked to AI.

83
00:07:02,480 --> 00:07:08,520
And we still need to, to have the technical knowledge, the technical background to ask the

84
00:07:08,520 --> 00:07:10,840
good classes to, to AI.

85
00:07:10,840 --> 00:07:18,160
So, we, we still need the programmers in, in the middle, or orchestrating agents, AI.

86
00:07:18,160 --> 00:07:29,520
But, uh, our, our work today is, is, is very, uh, it's, it's changes because, uh, I don't

87
00:07:29,520 --> 00:07:37,120
write, for example, uh, much more code as, as I read, uh, in the past, for example, this,

88
00:07:37,120 --> 00:07:44,160
this, this, few last months, for example, I wrote, uh, just a few lines of code by hand.

89
00:07:44,160 --> 00:07:49,360
Uh, most of my code is written by, AI is written by agents.

90
00:07:49,360 --> 00:07:57,120
And I can easily create Python scripts, uh, for, for, to, to automate my boring tests, but

91
00:07:57,120 --> 00:08:00,880
they, so it's crazy.

92
00:08:00,880 --> 00:08:10,480
It's, it's crazy how fast is, is changing and how it is, uh, generating good code day by day.

93
00:08:10,480 --> 00:08:14,080
And, and can you, what, what, what I, I found really interesting, what, what is the,

94
00:08:14,080 --> 00:08:19,200
the difference between AI coding assistant and AI coding agents?

95
00:08:19,200 --> 00:08:22,720
What I, what, yeah, that, that's nice.

96
00:08:22,720 --> 00:08:31,840
Uh, there, there are many approaches to, to reach, uh, the, the main goal for your software development.

97
00:08:31,840 --> 00:08:39,920
You can simply, uh, but, uh, put the AI assistants in your ID, for example, I was a source studio,

98
00:08:40,640 --> 00:08:50,240
from Microsoft. So I, I can simply open a tab, aside me, the studio and say, hey, what's going on here?

99
00:08:50,240 --> 00:08:56,720
What are suggestions for me to reflect or this method to implement this feature?

100
00:08:56,720 --> 00:09:03,920
So they, I can assist me. They can act as, uh, senior developer, um, aside me.

101
00:09:03,920 --> 00:09:13,280
And, uh, the, the good points is that I can say to AI, hey, uh, just focus on Microsoft documentation.

102
00:09:13,280 --> 00:09:19,520
Don't look anywhere around. Okay, just bring me answers from this source of truth,

103
00:09:19,520 --> 00:09:26,400
which is Microsoft learned, by the way, because they have, uh, the, the, the MCP, which was called,

104
00:09:26,400 --> 00:09:33,520
uh, the, the, the model context protocol, which is a protocol that allows us to, uh,

105
00:09:33,520 --> 00:09:43,200
uh, to scrape information inside, uh, Microsoft to learn website. So this is, uh, a way that AI can assist us.

106
00:09:43,200 --> 00:09:48,880
So we call this, uh, assistant AI or AI assisted code development.

107
00:09:48,880 --> 00:09:56,800
We, I think is the most common features nowadays. Uh, most people that I, I see programming use this,

108
00:09:56,800 --> 00:10:05,040
this kind of, uh, assistant. Uh, so we, we don't need to go to websites in more search for an answer.

109
00:10:05,040 --> 00:10:14,320
Uh, we can ask to, to AI inside our, uh, our ID, uh, and we can debate about this. So, uh,

110
00:10:14,320 --> 00:10:22,320
explain me why are you telling these? Explain me, uh, or show me three more ways to, to do this.

111
00:10:22,320 --> 00:10:33,440
So this is good. This is the, uh, assistant AI, the agent development is a little bit different because you can ask AI to do the complete test for you.

112
00:10:33,440 --> 00:10:46,160
By the way, you can ask AI to plan this before you even start. So you can simply open the same way, open, uh, the, the chat window and say,

113
00:10:46,640 --> 00:10:56,880
hey, uh, copilot, uh, by the way, I use copilot from, from Microsoft and it can connect to multiple, uh, AI agents.

114
00:10:56,880 --> 00:11:07,040
So cloud, uh, codex, uh, doesn't matter. I like it because it can talk to multiple agents. So say, hey, copilot,

115
00:11:07,040 --> 00:11:20,400
I need to do this task, uh, the, my problem is this, uh, what is our plan to do this? Then, uh, it can write a plan, uh, show me this plan,

116
00:11:20,400 --> 00:11:32,880
show me how he intend to implement this and then decide implementation. This is a, uh, a different approach because it can trigger multiple agents at the background.

117
00:11:32,880 --> 00:11:43,840
It split the task in sub-task, micro-task, trigger multiple agents. Uh, this is agents can run in different, uh,

118
00:11:43,840 --> 00:11:56,560
AIs like, codex, uh, uh, uh, cloud, uh, term nine and get the job known for me. So this is, this is by, by fire, one of the craze is way.

119
00:11:56,560 --> 00:12:17,760
And there one more way with this, the spec driven development, which we can, uh, write specifications and place this is specifications inside, uh, some specific folders inside our software and say, hey, base it on this, uh, specifications, uh, do the system for me.

120
00:12:17,760 --> 00:12:29,920
And, uh, sorry, and he starts to code it's, it's a little bit, uh, crazy, but it's, it's very fun and he's doing a very good job.

121
00:12:29,920 --> 00:12:39,520
Since you have a good specification, you have a good architecture, uh, very, well-defined, it can do, uh, a lot of work for you.

122
00:12:39,520 --> 00:12:49,440
Um, but, um, can I only prompt it or is there capabilities that do useful software development agent activity?

123
00:12:49,440 --> 00:12:57,520
Right. Uh, is what capabilities do, uh, useful software development agent activity need?

124
00:12:57,520 --> 00:13:25,520
No, it's just prompting. Just, you can just prompt it. It's, uh, uh, a natural language. Uh, so even if you, you don't have, uh, the, the, the specification, uh, yourself or even you're in the agent mode and stuck in a test for yourself, you can, uh, you have different modes to interact with.

125
00:13:25,520 --> 00:13:39,520
It's a component, for example, you have the agent, bold and the ask mode. So if you, if he's doing some task, you can interrupt him and, uh, use the ask mode to say, hey, uh, what is going on here?

126
00:13:39,520 --> 00:13:43,520
What are you implementing or why are you implementing this way?

127
00:13:43,520 --> 00:13:59,520
So we have a lot of flexibility during the, uh, the implementation. So this is, it's, it's very, very like, uh, a team working for you inside your, your ID.

128
00:13:59,520 --> 00:14:01,520
So that's fun.

129
00:14:01,520 --> 00:14:15,520
This is, this is awesome. Um, especially I only prompted so that's, that's cool. I don't need, I don't know some context or ask for awesome. I don't know.

130
00:14:15,520 --> 00:14:30,520
Yeah, that's, that's, that's, that's really, that sounds really simple. Like that. Yeah, yeah. Um, Microsoft has these, um, yeah, agent framework. Can you a little bit say what, what gives Microsoft agent framework is?

131
00:14:30,520 --> 00:14:44,520
Nice. That's nice. Uh, we, we talked to here about, uh, how we can code using agents, how we can delegate to, to, to agents, uh, our work, right.

132
00:14:44,520 --> 00:14:56,520
But, uh, this is just one part of our, our job. Other part is to integrate AI into our software. So I need to provide capabilities to my users.

133
00:14:56,520 --> 00:15:11,520
For example, to interact with AI or, uh, to create something we call Ragn, which is retrieval omitted generators to improve, uh, the quality of the response, uh, for our models.

134
00:15:11,520 --> 00:15:25,520
So there's a lot of stuff going on behind the AI and it's very difficult for us to manage all the things because we need to, uh, we talked about Ragn, MCP servers.

135
00:15:25,520 --> 00:15:43,520
There's a lot of stuff. So Microsoft agent framework is a tool, uh, is a set of tools that is built from Microsoft and, uh, it is provided us to, to do this interaction with AI, sign our software.

136
00:15:43,520 --> 00:16:07,520
It's very easy, for example, to, it's like plug and play. You can, uh, bring, uh, Microsoft agent AI to your software and connect with multiple AI's, which is, uh, is very common in the, it's, it's a little bit tricky to do because, uh, you talk different with, uh, codex and then you talk to, to call out for, for example.

137
00:16:07,520 --> 00:16:33,520
So Microsoft agent framework, uh, it creates a service API. It's created an interface for that. So you can talk directly to, to, to Microsoft agent framework and he talked to codex, he talked to cloud, he provide MCP servers, he provided, uh, uh, access to Rags, uh, he provided dating gesture, which is another process to, um,

138
00:16:33,520 --> 00:16:47,520
pick all the information that you have on your company and, uh, but it's either your database, which is, uh, seems to be, to be simple, but it is, is very complicated.

139
00:16:47,520 --> 00:17:12,520
The last part is that, uh, it's built by Microsoft. So if you work with .NET, for example, you have all the things, uh, ready for you for free. This is, uh, inside and you can speed up your, your development, your AI integration, your, your software, uh, within a click, for example, it's, it's very fun.

140
00:17:12,520 --> 00:17:25,520
So this is the, the other approach that we have to, uh, integrate AI into, into our software, which is, uh, I, uh, programming the school.

141
00:17:25,520 --> 00:17:45,520
And we, we did, we, we, we factor our, our system, the, the, the, the, this year, and we are integrating everything with AI. So, uh, people can, uh, during the, the, the lectures, people can ask something, say, hey, what is going on here?

142
00:17:45,520 --> 00:17:55,520
I did not understand this, this concept. So all this integration is, is made by, uh, Microsoft agent framework. So that's very cool.

143
00:17:55,520 --> 00:18:12,520
And I'm not, uh, that, that, that, that, that, that, that, that, that, that, that, that, that, that, that, that, what do is the agent framework give me that I don't already have by simply calling in their, I model through, through the app.

144
00:18:12,520 --> 00:18:37,520
No, it's more like, uh, a group of, of tools that you have, if you, if you were to, to call this by hand, you have a lot of, a lot of work to do because, uh, the way that we, we talk with models, uh, tend to be different from company company.

145
00:18:37,520 --> 00:18:57,520
For example, if you use, uh, Azure AI Foundry, which is awesome, uh, that you can concentrate on your, your models, your little bits ahead because, uh, Azure AI Foundry provide an API for you to, to call these models and, uh, it all pass through the same service.

146
00:18:57,520 --> 00:19:08,520
But as soon as you try to connect, for example, with Google to, to consume these models, you will have to call different API. You have to send different information.

147
00:19:08,520 --> 00:19:15,520
So, uh, you, you will need to deal with all this code inside your, your code base.

148
00:19:15,520 --> 00:19:19,520
So it's, it's a lot of code just to do simple tasks.

149
00:19:19,520 --> 00:19:27,520
So, math, math, as we call math, uh, it help us without the code.

150
00:19:27,520 --> 00:19:43,520
And, and, um, I think how, how nature is dot that act early for, yeah, for the, for these, in your generation of, of, of any application is, is it well positive?

151
00:19:43,520 --> 00:19:51,520
It's not so major, how good, good is dot, dot net, uh, that's a, programming language.

152
00:19:51,520 --> 00:19:54,520
That's a good, that's a good question.

153
00:19:54,520 --> 00:20:11,520
Because most people, uh, think that, uh, because that was created in back into, to thousand, uh, it is, uh, an old framework, dot net is the framework, C sharp is the language, the programming language.

154
00:20:11,520 --> 00:20:13,520
They were created together.

155
00:20:13,520 --> 00:20:21,520
And, uh, most people, uh, think that, uh, it's a little bit older, but it's not, C sharp is evolving.

156
00:20:21,520 --> 00:20:27,520
Now, we, we have major, uh, release from that every year in November.

157
00:20:27,520 --> 00:20:40,520
So, it, it's crazy because, uh, every year, they, they bring us new features, they bring us, uh, even more speed and, and, uh, and security side, dot net.

158
00:20:40,520 --> 00:20:42,520
Which is awesome.

159
00:20:42,520 --> 00:20:46,520
Uh, that's in, there's, uh, a very, very good job.

160
00:20:46,520 --> 00:21:07,520
Uh, and on the A.I. side, I have to confess that Python is, as the pioneer for, for AI and the, is, is the first choice because, uh, most of tools that we use today are based on, on, on Python, they were creating a Python, which is a very, uh, good language.

161
00:21:07,520 --> 00:21:12,520
Uh, especially for, for AI because it's very simple, it's very straightforward.

162
00:21:12,520 --> 00:21:26,520
Uh, but we, we have evolved so much with C sharp, uh, in a way that's, uh, C sharp, it's basically, uh, as, as easy to learn as, as Python over the day.

163
00:21:26,520 --> 00:21:41,520
You have a lot of features and with Microsoft agent framework, for example, uh, by the way, Microsoft agent framework is a, a, a, a, a, a, a, a, a, a, two previous product that Microsoft build with, which was cement, Carano, and auto-gain.

164
00:21:41,520 --> 00:21:57,520
So, when Carano was built to, uh, deal with models the way I, I told you, uh, and auto-gain was, uh, was a project that you, you can, uh, uh, orchestrate workflow, create pipelines using AI.

165
00:21:57,520 --> 00:22:15,520
So, Microsoft agent framework is, uh, is the combination of these two features. So, uh, is it, I can say that, that, uh, we have a complete workflow, uh, to work with AI sign.net, a complete and a modern workflow to work with AI sign.net.

166
00:22:15,520 --> 00:22:32,520
We still have more to come, uh, they are, they are planning more releases over Microsoft agent framework and, uh, I don't expect it to have the same popularity of Langchain, for example, which is, uh, the, the same as still, uh, but in Python.

167
00:22:32,520 --> 00:22:48,520
But, uh, I, I don't miss any, any feature, no, no, days, I can code and sign.net, I can use my, my code base. And by the way, not that is, uh, a very broad ecosystem in a very complex system.

168
00:22:48,520 --> 00:23:01,520
Uh, we can create mobile apps, game development, software development, AI, everything inside of that. So, it's very complex and it's very modern also.

169
00:23:01,520 --> 00:23:05,520
Awesome. Um, but how?

170
00:23:05,520 --> 00:23:19,520
I say, if an organization has years of business logic implemented in about that applications, how easily can that functionality become available to these AI agents?

171
00:23:19,520 --> 00:23:21,520
Good question.

172
00:23:21,520 --> 00:23:46,520
For example, uh, here we have a, our entire software is built inside that. So we have a lot of code is sharp here. And with math, for example, with just one line of code, we can export that method and create, uh, a metadata for that, uh, to AI communicate with that method.

173
00:23:46,520 --> 00:24:04,520
So, like an export a function of my system, to AI, I say that, uh, with just one line of code. So if you have dot that in, you start using math today, you can create an, an MCP server with a flick of the wrist.

174
00:24:04,520 --> 00:24:15,520
It's, it's very easy because it's C#, it's native code. So, uh, it's very easy to, to integrate, right?

175
00:24:15,520 --> 00:24:34,520
So that means, um, companies may not to, to rebuild their X, their existing application to become an AI application or a, I know, that they can start today, just, uh, bringing Microsoft agent framework, for example,

176
00:24:34,520 --> 00:24:46,520
we have other packages like MCP for, for that, which is, you can just add to your, your project and starting export MCP.

177
00:24:46,520 --> 00:25:05,520
You, you can start today, it doesn't need to, to plan anything, uh, much more complex, for example, you can start simple, uh, and then obviously, uh, to, to create an AI first, uh, on your company is, is, is something that involves,

178
00:25:05,520 --> 00:25:12,760
a lot of other things, but if you want to just integrate it with models to push

179
00:25:12,760 --> 00:25:17,060
AI to your application, it's very simple straightforward today.

180
00:25:17,060 --> 00:25:21,440
And, is there any architecture changes?

181
00:25:21,440 --> 00:25:28,720
So how do you build an AI application differ from building a, I say, traditional,

182
00:25:28,760 --> 00:25:39,320
domestic application? We tend to separate some sort of things because one thing

183
00:25:39,320 --> 00:25:45,240
that concerns me a lot using the AI is the secured, right?

184
00:25:45,240 --> 00:25:52,120
Because we don't want you to expose anything outside our, our organization,

185
00:25:52,120 --> 00:25:58,040
or for our users. And since they use natural language, it's, it's hard to

186
00:25:58,640 --> 00:26:04,680
to predict which kind of answer they will ask for our system.

187
00:26:04,680 --> 00:26:09,440
So we have to double check to create a lot of tasks.

188
00:26:09,440 --> 00:26:17,200
But depending on how your system is built, if you create small functions,

189
00:26:17,200 --> 00:26:20,600
if you pass these functions, if you ship these functions,

190
00:26:23,800 --> 00:26:30,480
granularly, it's, it's very easy to integrate with AI this way I told you,

191
00:26:30,480 --> 00:26:36,720
exporting these functions or wrapping these functions inside a, right?

192
00:26:36,720 --> 00:26:42,760
So usually we, we just create some separated class,

193
00:26:42,760 --> 00:26:51,320
just like I would practice to, to export this, to not export the core function.

194
00:26:51,880 --> 00:26:57,920
And in some cases, we create another project just for AI to, to separate the things,

195
00:26:57,920 --> 00:27:02,320
say just everything related to AI, you put the science project,

196
00:27:02,320 --> 00:27:08,440
but the architecture itself, you don't need to, to change anything in your

197
00:27:08,440 --> 00:27:15,360
architecture, you can export this. But if you're thinking to expand this to,

198
00:27:16,480 --> 00:27:23,320
to evolve this in a future, probably you need to rethink about your architecture,

199
00:27:23,320 --> 00:27:28,480
how you'll, you'll place AI inside your current software.

200
00:27:28,480 --> 00:27:31,320
And in your, your company in general.

201
00:27:31,320 --> 00:27:38,400
Yeah, also, I think when I think from, yeah,

202
00:27:38,400 --> 00:27:42,760
the architecture, I say tradition of software,

203
00:27:43,680 --> 00:27:49,320
follows logic like this X happened and execute J,

204
00:27:49,320 --> 00:27:51,960
which agent,

205
00:27:51,960 --> 00:27:58,880
all the, with an agent, we might be instead say, here's the goal that I'm

206
00:27:58,880 --> 00:28:06,560
in what actions are necessary. How profound is this, is this architectural change?

207
00:28:06,560 --> 00:28:10,280
Yeah, the nice, I got your question.

208
00:28:10,720 --> 00:28:13,720
It's a, it's a kind of a part of the shift, right?

209
00:28:13,720 --> 00:28:19,960
Because we, in software development, we tend to do things very strict.

210
00:28:19,960 --> 00:28:25,440
So you have one way, other way, you have to pass from a to b,

211
00:28:25,440 --> 00:28:27,800
this certain kind of information.

212
00:28:27,800 --> 00:28:33,520
And for example, exposing your, your, your functions,

213
00:28:33,520 --> 00:28:38,520
using Microsoft agent framework, for example, which will create an,

214
00:28:38,560 --> 00:28:44,200
an MCP server for, for your code, you can expose this function to AI.

215
00:28:44,200 --> 00:28:49,840
And you say basically, hey, this is the function to turn on the lights.

216
00:28:49,840 --> 00:28:53,400
Okay, this is the function to turn on the camera.

217
00:28:53,400 --> 00:28:57,400
This is the function to record some podcasts.

218
00:28:57,400 --> 00:29:03,320
Then with this information, you can use natural language to interact with it.

219
00:29:03,320 --> 00:29:07,200
And that, that is the, the cool part of AI,

220
00:29:07,720 --> 00:29:13,640
because we, we interact with AI today using natural language,

221
00:29:13,640 --> 00:29:19,280
but inside chat, GPT, inside a cloud, things like that.

222
00:29:19,280 --> 00:29:24,560
So we tend to, to do some prompt, this, but to write some text, to write some code.

223
00:29:24,560 --> 00:29:31,240
But no, they, we, we can create a prompt, for example,

224
00:29:31,240 --> 00:29:35,880
to, to create a product inside our database.

225
00:29:36,200 --> 00:29:43,040
So you can simply say, hey, register this product for me, the ID is one, the prices.

226
00:29:43,040 --> 00:29:46,280
There, and that's, that's fine.

227
00:29:46,280 --> 00:29:53,800
It's, if we expose the, the, the function to, to register a product, to create a product,

228
00:29:53,800 --> 00:29:57,160
we can interact with AI that way.

229
00:29:57,160 --> 00:30:02,640
Another cool thing is that if you're using a multi model, model,

230
00:30:03,520 --> 00:30:08,400
a multi model, model, you can simply take a picture, for example,

231
00:30:08,400 --> 00:30:12,360
from a product sent to that and say, register this product.

232
00:30:12,360 --> 00:30:18,200
It can bring information for you and try to register this product.

233
00:30:18,200 --> 00:30:24,440
So we stopped to, to do in some areas, right?

234
00:30:24,440 --> 00:30:28,480
We stopped to do all the methodic way that we used to.

235
00:30:28,480 --> 00:30:33,240
So you have the fields to feel you have the, all that stuff.

236
00:30:33,240 --> 00:30:38,560
On the screen and we can start to interact more with prompts with natural

237
00:30:38,560 --> 00:30:44,120
English say, hey, do this for me and it will find a way to do that.

238
00:30:44,120 --> 00:30:51,080
This, I classify this into two, two pieces.

239
00:30:51,920 --> 00:30:59,080
The first part is the IDA potent way, which must be that way it can change.

240
00:30:59,080 --> 00:31:03,720
You can, you should expect the same result always that, that seems.

241
00:31:03,720 --> 00:31:10,880
And then we have the, this new way that you can simply send a prompt to, to, to AI

242
00:31:10,880 --> 00:31:12,920
and that will follow the prompt.

243
00:31:12,920 --> 00:31:16,040
So we can interact with this way.

244
00:31:16,040 --> 00:31:19,440
That's really amazing.

245
00:31:19,440 --> 00:31:21,000
What's possible.

246
00:31:21,360 --> 00:31:26,680
So, yeah.

247
00:31:26,680 --> 00:31:36,240
What's happened was this traditional design patterns in, in, in an agent architecture,

248
00:31:36,240 --> 00:31:42,640
will they stay there or is it, is there, is completely new?

249
00:31:42,640 --> 00:31:50,560
No, I think the, the, the good practice of software development is still alive.

250
00:31:50,760 --> 00:31:56,160
So, right code one time is, is very easy to maintain, right?

251
00:31:56,160 --> 00:32:03,600
A maintainable code is very, is very hard as Uncle Bob says, because even AI,

252
00:32:03,600 --> 00:32:09,440
looking for a massive code will do a massive code, right?

253
00:32:09,440 --> 00:32:10,680
They follow the pattern.

254
00:32:10,680 --> 00:32:16,400
So if you say, hey, implement this, it will look to our code base and say, well,

255
00:32:16,400 --> 00:32:20,200
I implement this the way it is implemented on, on your code base.

256
00:32:20,600 --> 00:32:24,480
So we still need to think about patterns.

257
00:32:24,480 --> 00:32:28,560
We still think about to, uh, uh, uh,

258
00:32:28,560 --> 00:32:32,960
clean code, which is something we, we, we talk every day.

259
00:32:32,960 --> 00:32:39,320
But it, it's way easier to implement, uh, no days using AI.

260
00:32:39,320 --> 00:32:45,320
You can, uh, you can even forget about it for, for a second,

261
00:32:45,320 --> 00:32:49,320
when, when you're planning, for example, when I, I'm planning, uh,

262
00:32:49,320 --> 00:32:53,000
I, I don't specify any design pattern for AI.

263
00:32:53,000 --> 00:33:00,200
I, I, I live a blank canvas at least for, for the beginning and say, uh,

264
00:33:00,200 --> 00:33:01,400
let's do this plan.

265
00:33:01,400 --> 00:33:03,360
They will come with a plan for me.

266
00:33:03,360 --> 00:33:09,000
And, uh, if the implementation is way different than, then I think you say,

267
00:33:09,000 --> 00:33:12,720
hey, but what about using the strategy pattern here?

268
00:33:12,720 --> 00:33:16,600
What about using the prototype design pattern here?

269
00:33:17,080 --> 00:33:22,040
And I, I can suggest you, to, to then, uh, and, and do the,

270
00:33:22,040 --> 00:33:25,560
this kind of implementation because it's, uh, create some,

271
00:33:25,560 --> 00:33:32,800
maintainable code, uh, two, two things that, uh, you must achieve in software development

272
00:33:32,800 --> 00:33:38,760
is a testable code because, uh, when everything changes and, uh,

273
00:33:38,760 --> 00:33:42,480
everything will change, uh, you need to pass everything.

274
00:33:42,480 --> 00:33:46,800
And I'm good looking and maintainable code, right?

275
00:33:46,800 --> 00:33:50,200
So that these two, two things are negotiable.

276
00:33:50,200 --> 00:33:54,080
Uh, you, you always need to achieve this, just two things.

277
00:33:54,080 --> 00:33:58,920
So, uh, design patterns, best practices, still present.

278
00:33:58,920 --> 00:34:04,440
Uh, they, they, a little bits, uh, on the depth layer now because AI,

279
00:34:04,440 --> 00:34:11,400
we interact more with the surface, uh, the, the code itself, but, but they need to be

280
00:34:11,400 --> 00:34:11,800
there.

281
00:34:11,800 --> 00:34:14,400
Uh, you, you, you, you still need them.

282
00:34:15,080 --> 00:34:15,920
Yeah.

283
00:34:15,920 --> 00:34:22,040
Yeah, I think one of the biggest conceptual change is that developers are used to,

284
00:34:22,040 --> 00:34:26,600
yeah, deterministic systems and AI models are probabilistic.

285
00:34:26,600 --> 00:34:33,920
So, um, how difficult is this transition, especially when I think, um,

286
00:34:33,920 --> 00:34:38,040
when you solve, when you test software and, uh, yeah,

287
00:34:38,040 --> 00:34:43,400
the same input, not necessarily, necessarily produces the exact same output.

288
00:34:44,320 --> 00:34:45,160
Okay.

289
00:34:45,160 --> 00:34:45,360
Nice.

290
00:34:45,360 --> 00:34:53,360
Uh, we call this stochastic because, uh, every, every AI generation is, is different from the,

291
00:34:53,360 --> 00:34:54,240
the previous one.

292
00:34:54,240 --> 00:35:03,760
So, uh, that, this is, I think this is the, uh, the most, uh, difficult part of interacting

293
00:35:03,760 --> 00:35:10,280
with AI being a programmer because most of, uh, programmers that I, I, I know today,

294
00:35:10,320 --> 00:35:13,960
expect the same result because we were built to that.

295
00:35:13,960 --> 00:35:19,600
So, uh, we, we do this when they expect the, the same, the same result, but this is not,

296
00:35:19,600 --> 00:35:23,920
uh, this won't happen with AI even generating a code.

297
00:35:23,920 --> 00:35:28,680
If you ask the, the same code generation to, twice in a, in a minute,

298
00:35:28,680 --> 00:35:31,640
it will generate complete, different codes.

299
00:35:31,640 --> 00:35:37,280
So we have to, to, to keep that in mind when we're at your specifications.

300
00:35:37,640 --> 00:35:46,080
So we, we are focused, we're focusing more on algorithms on, uh, how things should work rather

301
00:35:46,080 --> 00:35:47,560
than how the code is written.

302
00:35:47,560 --> 00:35:55,600
So, uh, basically what I try to do is, uh, understand that, uh, there are many ways to do the

303
00:35:55,600 --> 00:35:56,120
same thing.

304
00:35:56,120 --> 00:36:05,720
So, I don't have to, uh, to do that, uh, the way I intend to do, I can leave it to, to, to,

305
00:36:05,720 --> 00:36:08,960
away, I, I just have, I, I just need to have this don't.

306
00:36:08,960 --> 00:36:18,320
So, I don't, I don't try to, to, uh, get to read to my, to my code to, to, to,

307
00:36:18,320 --> 00:36:22,720
the, the generated code I prefer to, uh, line it to, to, to,

308
00:36:22,720 --> 00:36:25,160
A.I. Just, uh, validate this code.

309
00:36:25,160 --> 00:36:27,320
If this, this code is, is good.

310
00:36:27,320 --> 00:36:33,280
If, if, if he, he written the, uh, I, I go to approach, uh, I, I don't mind.

311
00:36:33,760 --> 00:36:35,680
So, that, that's fine.

312
00:36:35,680 --> 00:36:37,440
Let's, let's do this way.

313
00:36:37,440 --> 00:36:43,640
It's like, uh, you're, you're delegating the, the, the creation to you, another person, uh, barely,

314
00:36:43,640 --> 00:36:49,160
uh, it's very hard to find a, uh, another person who, which will code like you code.

315
00:36:49,160 --> 00:36:53,000
So it, it will code different from you.

316
00:36:53,000 --> 00:36:59,840
Uh, they, both of you can, can achieve the same result, but, uh, the, the approach will,

317
00:36:59,840 --> 00:37:00,720
will be different.

318
00:37:00,720 --> 00:37:03,040
So, uh, using the, uh, is that way.

319
00:37:03,320 --> 00:37:10,560
I think the, the most challenging, uh, things behind it, uh, is more to, to test the software.

320
00:37:10,560 --> 00:37:18,320
Because when we interact with, with, with A.I., for example, uh, we have millions, uh,

321
00:37:18,320 --> 00:37:20,680
uh, ways to, to do the things.

322
00:37:20,680 --> 00:37:25,080
So, uh, when, when you expose this to, to million,

323
00:37:25,080 --> 00:37:30,000
user, you have million users with medium ways to do the, the, the, the same thing.

324
00:37:30,000 --> 00:37:35,560
So you need to, to create a very good, hard, award, rails, uh, to, to that.

325
00:37:35,560 --> 00:37:41,840
We had so, so many cases of prompt injection, which is a, a technique to, to,

326
00:37:41,840 --> 00:37:48,280
grabbing information and to, uh, hack your, your system using A.I.

327
00:37:48,280 --> 00:37:52,280
So I think this is the most cool, certain part.

328
00:37:52,280 --> 00:37:59,200
And, and as, as you may imagine, uh, it's very hard to predict everything that, uh, people will say,

329
00:37:59,680 --> 00:38:07,440
uh, there, there was a, uh, a fun case with, uh, a guy, uh, asked the A.I.

330
00:38:07,440 --> 00:38:11,760
to sing, uh, some Windows keys for, for him.

331
00:38:11,760 --> 00:38:14,640
So, he told A.I.

332
00:38:14,640 --> 00:38:24,280
a story that he was, uh, his grammar, sing, uh, song, uh, some Windows keys for him when he,

333
00:38:24,280 --> 00:38:26,080
he was young and A.I.

334
00:38:26,080 --> 00:38:28,520
gave him some, or he told me those keys.

335
00:38:28,960 --> 00:38:36,200
So, uh, that, that was for, but, uh, it was for us, but not so fun for the, the company running.

336
00:38:36,200 --> 00:38:40,720
They, they, I, so, but we, we have to, to be aware of that.

337
00:38:40,720 --> 00:38:48,240
So if you're exposing the, uh, a point for the, the people to, uh, write something inside our

338
00:38:48,240 --> 00:38:53,920
system to interact with, with our system, you, you, you should have the, the, the guard array is in,

339
00:38:53,920 --> 00:38:54,920
in, in, in mine.

340
00:38:55,280 --> 00:39:01,240
And if you're coding is the, the same way, if you're coding, if you're asking A.I.

341
00:39:01,240 --> 00:39:07,680
to write your code, you have to, you need to have the, the guard array is to say, Hey, don't do that.

342
00:39:07,680 --> 00:39:11,840
Uh, try to implement this pattern, try to do this way.

343
00:39:11,840 --> 00:39:13,120
Right.

344
00:39:13,120 --> 00:39:14,440
Yeah.

345
00:39:14,440 --> 00:39:20,840
Well, when I look a little bit into the, the developers groups are in, in, in, in YouTube, I have the feeling,

346
00:39:21,280 --> 00:39:31,600
uh, that's often focused on, on, on the model to use, uh, but the, how important is this, is context compared,

347
00:39:31,600 --> 00:39:34,840
uh, with, with model selection from your perspective.

348
00:39:34,840 --> 00:39:44,200
Uh, I think the context for A.I. is everything, you know, we, we need to, uh, understand and manage your

349
00:39:44,200 --> 00:39:53,040
content, context, uh, uh, very carefully, because, uh, if you, if you write a simple context, A.I.

350
00:39:53,040 --> 00:39:57,360
Won't do the, the job and few write a complex, uh, context, A.I.

351
00:39:57,360 --> 00:39:58,520
We'll lose itself.

352
00:39:58,520 --> 00:40:05,360
So, yeah, you need to, to write, uh, the, just, to stuck in the middle of this.

353
00:40:05,360 --> 00:40:12,360
And sometimes it's very hard because sometimes we think we know about the context and we, we, we, we, we,

354
00:40:12,360 --> 00:40:21,640
doesn't know. So as, as a programmer, what, what I usually do is basically, uh, implement some business rules.

355
00:40:21,640 --> 00:40:27,600
So if I don't know the business rules, how can I ask for a item, implement for me?

356
00:40:27,600 --> 00:40:29,200
How can I implement myself?

357
00:40:29,200 --> 00:40:38,080
So more than never, uh, we, we know they need to get into the business.

358
00:40:38,160 --> 00:40:45,200
So we need to understand the business, understand the needs and another thing that I, I, I, I always, uh,

359
00:40:45,200 --> 00:40:52,640
talk to my students, for example, uh, you need to question, you need to, to ask questions to your,

360
00:40:52,640 --> 00:40:56,280
your business and say, hey, what is going to hear?

361
00:40:56,280 --> 00:40:57,560
What are implementing?

362
00:40:57,560 --> 00:41:00,120
Why are we implementing here?

363
00:41:00,120 --> 00:41:06,760
Because, uh, with this question in, in mind, we can, uh, provide this to, to A.I.

364
00:41:06,800 --> 00:41:09,760
into, into have a, uh, good context.

365
00:41:09,760 --> 00:41:19,680
And this is the hardest part to, to, to be true because, uh, software development is, is a little bit straightforward.

366
00:41:19,680 --> 00:41:27,920
So you have the good practices, you have, uh, the, the concept, the base content, the fundamental concepts of computing,

367
00:41:27,920 --> 00:41:36,520
um, which sends, uh, CSI, I was born and from, from now, but, uh, the context itself is,

368
00:41:37,160 --> 00:41:38,080
it, it's hard.

369
00:41:38,080 --> 00:41:44,800
When I, I talk to you, you, you have something in mind, you say to me that when I talk to another person,

370
00:41:44,800 --> 00:41:49,800
they have a little bit different context, a little bit, uh, point of view.

371
00:41:49,800 --> 00:41:55,120
So merge, uh, everything of that and pass this to, to, to A.I.

372
00:41:55,120 --> 00:41:57,400
It's a, it's a, a very, very challenge.

373
00:41:57,400 --> 00:42:04,920
And you have to mix that this, this business, uh, context with the technology context, because A.I.

374
00:42:04,920 --> 00:42:08,360
also lead, uh, the directions on the technical way.

375
00:42:08,360 --> 00:42:17,200
So which architecture we are using, which stack, we, we are using, uh, what comes next for, for, for our software.

376
00:42:17,200 --> 00:42:23,360
And so, but, uh, all this together is, uh, is a, is a very, is very challenging.

377
00:42:23,360 --> 00:42:29,360
So, we are, we are still learning from, from that because A.I.

378
00:42:29,440 --> 00:42:35,000
is relatively new. We have a few years, but, uh, it's relatively new.

379
00:42:35,000 --> 00:42:41,440
Yeah, I think when we talk about context, we have also talked about retrieval, augmented,

380
00:42:41,440 --> 00:42:51,120
generation, right, um, our agents, essential turning rack from receive information and

381
00:42:51,120 --> 00:42:54,160
answer into receive information decided act.

382
00:42:55,640 --> 00:43:02,480
Yeah, sort of, uh, uh, uh, rugged itself is not just built for, for A.I.

383
00:43:02,480 --> 00:43:11,360
And, uh, I think, uh, most of, uh, companies that are doing AI first, uh, the first thing I recommend is to create a good

384
00:43:11,360 --> 00:43:17,560
rag because create a rag is easy, create a good rag is, is, is, is other.

385
00:43:17,560 --> 00:43:24,320
It's a little bit complicated because, uh, is, is not just, uh, pool information inside.

386
00:43:24,320 --> 00:43:28,400
You have to, uh, to verify this information.

387
00:43:28,400 --> 00:43:31,960
You have to cut the blocks of this, this information.

388
00:43:31,960 --> 00:43:39,400
For example, uh, big challenge we had this, this year was to, uh, take all lessons that we,

389
00:43:39,400 --> 00:43:46,240
we have on our platform, it transcribe all these lessons and for each, uh, phrase that, that we have,

390
00:43:46,240 --> 00:43:53,840
we create an, uh, an entry point inside our, our, our, so the user can search for a specific word and go

391
00:43:53,840 --> 00:44:04,280
to straight to that point at the, the video, uh, within that, that we have this, this day, so it can be, uh, very, very complex.

392
00:44:04,280 --> 00:44:11,160
And at, at the other hand, it provides a lot of context, uh, to, to AI.

393
00:44:11,160 --> 00:44:21,040
So when you prompt something, uh, you, you can, you just can achieve a good response to, to our customer when you, you have data.

394
00:44:21,680 --> 00:44:31,800
Uh, the, the core, uh, behind AI is the data, the data you capture, the data you, you trade, the data, you, you pull, uh, into your, your, your

395
00:44:31,800 --> 00:44:33,560
existing, your, your database.

396
00:44:33,560 --> 00:44:36,040
Uh, this is the most important thing.

397
00:44:36,040 --> 00:44:44,280
And, uh, when you plug this together, uh, rag is the way that you, you have to, to provide this, this, this information.

398
00:44:44,280 --> 00:44:51,320
So you have to, to make the right cuts on this data, uh, to, to, to, to do this context.

399
00:44:52,280 --> 00:44:54,560
We all know that we have a context swing though.

400
00:44:54,560 --> 00:45:08,200
So the context is, is limited to, to, to an amount of torque for, for example, and we, we also know that, uh, there's a sweet spot on the, the

401
00:45:08,200 --> 00:45:13,000
context is, uh, two thousand, uh, two hundred thousand, uh, tokens.

402
00:45:13,000 --> 00:45:17,960
For example, even the context, uh, the, the model have the capability to one medium tokens.

403
00:45:18,360 --> 00:45:32,440
We know that, uh, it starts to, to, to, to, to, to, uh, so, uh, we need to do the right cuts on, on our data to provide the, the, the right amount of data when, uh,

404
00:45:32,440 --> 00:45:34,360
interactive with, uh, rag.

405
00:45:34,360 --> 00:45:46,920
So we pick up the, uh, for example, an entire article with, uh, two thousand words, but if you inject two thousand words to, to our context, uh, it, it can

406
00:45:47,760 --> 00:45:53,160
be bad instead of good because it's a lot of information that it doesn't need.

407
00:45:53,160 --> 00:46:04,560
So you have to chunk this, this data, put it into, uh, into a rag, for example, and when the user, uh, asked something, you can just go to your, your, your,

408
00:46:04,560 --> 00:46:12,320
your rag, pick just that paragraph and send it back to, yeah, so this is the importance of ragging.

409
00:46:12,320 --> 00:46:19,880
This is how it improves the, the, the, the context and the, uh, the, the answer to your customer.

410
00:46:19,880 --> 00:46:30,280
I think one of the, uh, candidate rack have, it's, it's the memory in that coming, I don't know, new, new ideas like Vicky, LMM,

411
00:46:30,280 --> 00:46:36,600
I don't think he, I think that's the, what, what should develop us think about the memory topic?

412
00:46:36,600 --> 00:46:38,280
Yeah.

413
00:46:38,960 --> 00:46:51,080
The memory itself is, uh, is the trickiest part when you say, uh, the context, uh, we, we are talking, uh, to the, the, the,

414
00:46:51,080 --> 00:47:04,600
talking about the memory itself, uh, you need to understand, uh, a little bit about, uh, how AI works, for example, uh, because this is, uh,

415
00:47:04,640 --> 00:47:12,720
this is the, the fundamentals of, of AI and this can help you to gather better answers from AI.

416
00:47:12,720 --> 00:47:22,800
For example, when you start a chat with Copilot, for example, and you say, Hey, Copilot, uh, do implement this for me.

417
00:47:22,800 --> 00:47:23,400
Okay.

418
00:47:23,400 --> 00:47:26,800
It will try to read your code base.

419
00:47:26,800 --> 00:47:28,280
It's right to read your code.

420
00:47:28,600 --> 00:47:36,440
But maybe, uh, a lot of code that is, that is your code base is meaningful for, for that action.

421
00:47:36,440 --> 00:47:44,240
So it will update its memory for that chat with a lot of, uh, a useful information.

422
00:47:44,240 --> 00:47:54,880
Instead of this, you can say, uh, Hey, Copilot, can you implement this code for me, base it on, and you can specify this fire,

423
00:47:54,880 --> 00:47:56,120
disfile this fire.

424
00:47:56,120 --> 00:48:08,880
So you, you gave Copilot, uh, uh, a, a, a very, a very good context with examples, with simple and straightforward, uh, example, with good example, because I,

425
00:48:08,880 --> 00:48:19,840
I, I, I, up to your, your entire code basis is written on the, uh, at the same way and the same, uh, uh, uh, the same way.

426
00:48:20,080 --> 00:48:25,840
So there, there are a piece of code that you wrote with more passion than, than others.

427
00:48:25,840 --> 00:48:36,240
So you can point this to, to, to Copilot, for example, uh, and all of this information when you start a chat is going to a memory is going to,

428
00:48:36,240 --> 00:48:39,080
to the context itself and it have a limit.

429
00:48:39,080 --> 00:48:47,640
So what's the average hat we do is to compact this conversation since you reach, uh,

430
00:48:47,920 --> 00:48:51,840
on a certain amount of, of tokens reach it there.

431
00:48:51,840 --> 00:49:02,000
And this kind of, the, the trick is part because when, uh, Copilot, uh, for example, compacted by itself, it will just do, uh,

432
00:49:02,000 --> 00:49:08,600
compact, it will say, Hey, uh, compact is for me or, uh, summarize this for me.

433
00:49:08,600 --> 00:49:09,520
That's fine.

434
00:49:09,520 --> 00:49:10,360
This is okay.

435
00:49:10,360 --> 00:49:13,760
It can work, but you can compact it to yourself.

436
00:49:13,880 --> 00:49:19,640
If the conversation is, it's getting a bit longer. You say, Hey, compact this conversation.

437
00:49:19,640 --> 00:49:23,800
Uh, the main points here are these, these, these, and this.

438
00:49:23,800 --> 00:49:34,040
So you'll have a, a, a very different kind of, of memory update at that point because you point and then to, to,

439
00:49:34,040 --> 00:49:38,080
to, the, the important topics of, of this conversation.

440
00:49:38,520 --> 00:49:47,560
And another thing that I do is every, every time that I finish, uh, something that I say, I achieve that,

441
00:49:47,560 --> 00:49:58,640
that what I was looking for, uh, I close the conversation and start another conversation because it starts with a clean context that,

442
00:49:58,640 --> 00:50:05,680
that would be say, you can clear the context, but I prefer to start a new, a new context.

443
00:50:06,320 --> 00:50:16,240
And, uh, regarding memory, uh, we can also have a, a, Markdown files, uh, with this, the summaries that I, I told you,

444
00:50:16,240 --> 00:50:18,120
inside our folders.

445
00:50:18,120 --> 00:50:26,320
So you, you, you can, uh, summarize our conversation with your specification and other things, place inside a, a folder.

446
00:50:26,320 --> 00:50:35,480
And every time that copilot reads that, that, uh, I'm saying copilot, but it's the same behavior for, uh, in AI that, that, that you use.

447
00:50:35,920 --> 00:50:44,600
So it, it will reach this, this Markdown file inside our folder and update its memory with, uh, with all the, this stuff.

448
00:50:44,600 --> 00:50:55,960
So, uh, this is cool, but you, you need to understand how, uh, uh, a lot of works, how the memory works because this is, is very important for every human being.

449
00:50:55,960 --> 00:50:57,360
Not as,

450
00:50:59,720 --> 00:51:12,720
uh, I think in a, not a topic, when we talk, especially AI, uh, agents, a company, company is really, yeah, enthusiastic on the one part, but on the other part, there's the security part.

451
00:51:12,720 --> 00:51:16,960
And, yeah, um,

452
00:51:16,960 --> 00:51:28,840
And I also see it in a lot of companies, they say, okay, we make it and then, uh, we have to train it, uh, and, and then they give, I don't know, access to reach.

453
00:51:29,600 --> 00:51:34,400
Uh, rights, uh, the world exchange show us and so on.

454
00:51:34,400 --> 00:51:36,120
We have to train it and so on.

455
00:51:36,120 --> 00:51:53,720
Um, what, what did you, you know, shoulder for me for the start, I think a little bit, we should give, uh, agents, the same principle of last privilege, like we do it for humans, or, uh,

456
00:51:53,720 --> 00:51:56,760
is it, uh, how should we think?

457
00:51:56,760 --> 00:51:57,720
Yeah.

458
00:51:59,000 --> 00:52:12,280
Uh, it, I'm a little bit concerned with security and we, we had so many leaks, cases, data leaks, cases in the last few years.

459
00:52:12,280 --> 00:52:21,280
So, uh, I tend to close all the doors and just open the door that you really need to.

460
00:52:22,040 --> 00:52:39,120
And I think, uh, Azure AI founder is doing a very good job for that because you can, uh, simply plug the models, you can supply, uh, create an AI infrastructure for, for your, for your scenario.

461
00:52:39,120 --> 00:52:50,000
And you can pass by a, best use in the Android ID, for example, which is the Microsoft, uh, looking at the previos active director.

462
00:52:50,360 --> 00:52:57,520
So, uh, you can write the permissions, uh, on, on the portal is, is very easy.

463
00:52:57,520 --> 00:53:05,920
I think it is a very good approach to have a security layer, uh, in front of your AI models.

464
00:53:05,920 --> 00:53:19,600
This is the, I think the main thing you, you need to do it, uh, the, the very first thing you need to do because we are not just concerned about the users outside, but the users inside our company, right?

465
00:53:19,640 --> 00:53:23,600
Because, uh, it can, can be massive.

466
00:53:23,600 --> 00:53:25,400
Someone can ask for this.

467
00:53:25,400 --> 00:53:32,360
Hey, give me the salary base from, from my colleagues and issue.

468
00:53:32,360 --> 00:53:36,600
You want to, you don't want to, to have this, this problem.

469
00:53:36,600 --> 00:53:47,480
Uh, and inside our applications, we also need to, to create guard, raise, uh, for, for that, for example, uh,

470
00:53:47,480 --> 00:53:52,240
I told you that we can expose some functions, uh, inside our system, right?

471
00:53:52,240 --> 00:53:53,840
And that is true.

472
00:53:53,840 --> 00:54:02,440
And that is one, you, you can, with one line of code, expose, uh, uh, uh, a function, for example, to list our products, right?

473
00:54:02,440 --> 00:54:13,240
But, uh, let's say that is some user interact with that and say, hey, bring me all products, including, uh, the ones that were not launched yet.

474
00:54:13,760 --> 00:54:24,320
So you have a problem because if your function, uh, are exposing the products, which was not launched yet, you have, uh, a leak on your, on your code.

475
00:54:24,320 --> 00:54:42,760
So, in some, in some cases, we need to put some, some, uh, wide layers on this, this function is a, uh, instead of exporting this function, let's create a new function, which exposed just the products, uh, itself,

476
00:54:43,200 --> 00:54:57,280
without, uh, uh, release date, for example, uh, does, does the release it products and, uh, let's expose this function instead of the function that's supposed all the, the, the products, uh, for the customers.

477
00:54:57,280 --> 00:55:03,840
So as we discussed previously here, uh, this is the item, the way.

478
00:55:03,840 --> 00:55:13,160
So every time that you ask for the products, you got, you get the products, there's a matter of which prompt you, you write, you won't get the launch and product.

479
00:55:13,160 --> 00:55:16,560
So that, that's the way we have to, to leave it.

480
00:55:16,560 --> 00:55:21,400
So I, I don't like to expose everything to, to AI.

481
00:55:21,400 --> 00:55:37,120
I prefer to create small functions for, uh, the, the money situations and to, uh, and expose these functions because I can control inside the code, which, uh, which can be done inside my, my code base.

482
00:55:37,120 --> 00:55:38,200
Right.

483
00:55:38,200 --> 00:55:39,920
Uh, uh,

484
00:55:41,000 --> 00:55:55,880
so it's the same principle to, when you connect things because, uh, I know that it is, is, is very fun and, and very straightforward to, uh, to connect to your AI to your SQL database, for example.

485
00:55:55,880 --> 00:55:57,840
It's very powerful.

486
00:55:57,840 --> 00:56:08,480
Uh, you go, you connect to your SQL database and say, Hey, this is my database and, and now, uh, let's do some, some magic here.

487
00:56:08,480 --> 00:56:12,200
Bring me this information that, that's fine.

488
00:56:12,200 --> 00:56:13,680
That's, that, that's cool.

489
00:56:13,680 --> 00:56:19,800
If that's, if you, just, if you're working alone, for example, that's very fine.

490
00:56:19,800 --> 00:56:36,040
But if you work with multiple people, uh, this is, uh, this is likely, uh, insane because you're given access to your entire database or to, for an entire table, uh, for people asking whatever they want.

491
00:56:36,040 --> 00:56:42,760
So they, they can ask if I want to build a better product for days and start a new component.

492
00:56:42,760 --> 00:56:43,840
How should I do?

493
00:56:43,840 --> 00:56:44,720
Okay.

494
00:56:44,720 --> 00:56:46,680
You have access to all the information.

495
00:56:46,680 --> 00:56:48,640
Data is the new oil.

496
00:56:48,640 --> 00:57:04,240
I think what, what, or the older topic, uh, a lot of companies talk about who was the, the topic prompt injection, but it's especially with MCPs, uh, the, the, the, the RAC pull attacks.

497
00:57:04,240 --> 00:57:15,400
Uh, the new, new chapter, can you a little bit explain how, how to prevent and I think for the people, they don't understand what, what, what, what's the different.

498
00:57:15,400 --> 00:57:16,640
Wow.

499
00:57:16,640 --> 00:57:17,440
Nice.

500
00:57:17,440 --> 00:57:28,000
Uh, property injection is the way that we, since we have an entry point that we can, uh, talk to AI models, we can ask whatever we want.

501
00:57:28,000 --> 00:57:30,520
So there is no limitation for that.

502
00:57:30,520 --> 00:57:31,840
Uh,

503
00:57:32,840 --> 00:57:39,000
And then, uh, since we have no limitations, we are open to the people creativity.

504
00:57:39,000 --> 00:57:41,600
So, and they are very creative.

505
00:57:41,600 --> 00:57:44,680
Uh, they can ask everything about your system.

506
00:57:44,680 --> 00:57:51,320
So if, if you don't, do not have any kind of grad radius for, for that, uh, it will expose data.

507
00:57:51,320 --> 00:57:56,280
It will even worth let people write data inside your, your system.

508
00:57:56,280 --> 00:58:10,000
So we have, uh, so many cases that people were using, uh, a chat box around the word to, to, to, uh, solve programming questions.

509
00:58:10,000 --> 00:58:24,360
For example, uh, instead of paying for a clause to, to ask, to ask my questions, I can go to a famous, uh, website out there, won't say the name because of a copy writing.

510
00:58:24,920 --> 00:58:37,320
Uh, but I can go to the website, start a chat, a support chat and ask the same questions and how can I, uh, do this function in Python and the chat will, will answer.

511
00:58:37,320 --> 00:58:39,360
It's, it's a very simple thing.

512
00:58:39,360 --> 00:58:43,240
It's the same, but it, it's, uh, spend your tokens.

513
00:58:43,240 --> 00:58:44,040
Yeah.

514
00:58:44,040 --> 00:58:48,280
It's, it's, it's costing for, for your component to transfer that question.

515
00:58:48,280 --> 00:59:09,480
It, it's the, the, uh, prompt injection is a little bit like, uh, in the old times where I can write into, uh, that forms, uh, um, database code and it takes it, uh, and say, okay, uh, I'm, uh, I'm, uh, I'm, I'm, I'm, I'm, um, and, uh, I was on the database.

516
00:59:09,480 --> 00:59:16,920
It's, it's all that, like, like this, but what, what, what is the, what's the right pull attack the people now?

517
00:59:16,920 --> 00:59:21,720
Yes, it's the same, the, the term that you're, you're saying is the SQL injection.

518
00:59:21,720 --> 00:59:28,160
We, we can, uh, write, uh, some, some prompt is, uh, this is the same.

519
00:59:28,160 --> 00:59:29,400
It's basically the same.

520
00:59:29,400 --> 00:59:43,360
So now we, we can, instead of injecting SQL, we can inject prompts and looks through the Trilett database, uh, ask questions about the things that is not, uh, in that context.

521
00:59:43,360 --> 01:00:06,080
So, uh, to, to prevent that, the, uh, I really recommend you to, to use something like, uh, Azure AI Foundry, for example, if you're doing a very, uh, complex, uh, or, um, AI first thing, uh, because it, it brings you, uh, a lot of Guadrass, do that.

522
01:00:06,080 --> 01:00:29,040
Uh, when you, you create a, when you start interacting with a model, you can create some, uh, system prompts, which we, we call it some kind of some sort of memory that you have that you can say, for example, do not answer questions about and say some talk.

523
01:00:29,040 --> 01:00:43,200
Even, uh, when you customer, every time that customer asks, uh, about something that is not related to hamburgers or to some sort of products, do not answer this, this question.

524
01:00:43,200 --> 01:00:57,880
So you can specify this, uh, when I say, create a word array, it seems to be something very complex is not, it's just a pure text that you, you can write a sign or a model that you can, uh, bring sign or model.

525
01:00:57,880 --> 01:01:10,840
But it's, it's hard to predict because, uh, it's, it's very hard to, to, to predict what users can, can think and can ask you to your model.

526
01:01:10,840 --> 01:01:25,880
So if you do not have anything, uh, prepared or anything, uh, how can I say, uh, trade it to, to, to, to that.

527
01:01:25,880 --> 01:01:35,880
Uh, you will, you end up having to write so many business rules that it's not, uh, it's not, it's painful.

528
01:01:35,880 --> 01:01:41,880
So, uh, I recommend you to, to, to bring, uh, something like Azure AI founder.

529
01:01:41,880 --> 01:01:53,880
I think I said, but basically Azure here because it's them, uh, the two that I, that I work, but there are many tools outside, you don't have to, to be strong with, with Azure, for example.

530
01:01:53,880 --> 01:02:07,880
But you bring something, something closer to your, to your business, uh, with business rules, uh, or an inside it, uh, with blocking your context, for example, uh, to prevent that.

531
01:02:07,880 --> 01:02:17,880
So, um, we have all, all, all these, these cool things we can now do with, with AI, especially encoding.

532
01:02:17,880 --> 01:02:27,880
Um, but, but I think, um, yeah, we have this, the tokens and the tokens are actually about getting cheaper for my feelings.

533
01:02:27,880 --> 01:02:46,880
Um, is it a show, developers, uh, also focused on architecture and the look, especially, I think fine ops can be find a sheep, a solution, or is, is it, uh, the time invest for the sheep, a solution.

534
01:02:46,880 --> 01:03:09,880
It makes it sense. So I, I don't know, I, I, I can use, uh, AI to, uh, uh, make me, uh, PF to document or show that use OCR because it's cheaper, like, something like this, as example, is it, what would you think?

535
01:03:09,880 --> 01:03:22,880
Yeah, I think and I, uh, and I use here is, is a mix of things, uh, AI is, uh, is a, is a broad term and it's been here since 1996.

536
01:03:22,880 --> 01:03:28,880
So it is, it's nothing new, but what we have new here is the generative AI.

537
01:03:28,880 --> 01:03:32,880
So the words say itself is generative, is to generate.

538
01:03:32,880 --> 01:03:43,880
So every time that I need to generate something new, I delegate this to AI, this is clear for me. If I need to do a repat it, task, this is not AI for me.

539
01:03:43,880 --> 01:03:59,880
For example, uh, every video that I do for YouTube, I have to extract the audio to do a transcription of this, this audio, uh, generate the, uh, the SR, SRT files to upload to YouTube.

540
01:03:59,880 --> 01:04:09,880
And translate this SRT to other languages, so I have the, uh, I have the, the translations, uh, on, on YouTube.

541
01:04:09,880 --> 01:04:26,880
So this is a process. This is a workflow. So I can send my video to Claude, for example, say, hey, do the transcription, uh, create the SRT portuguese, uh, anguish and Spanish for me.

542
01:04:26,880 --> 01:04:39,880
It will, it will do a very good job. It's, it's fine. It's pretty, it's fine. But it will, it will cost me a lot of tokens because it will do the same process that I do locally.

543
01:04:39,880 --> 01:04:48,880
So every time that I've, that I get a, uh, a video file here, I run a Python script, which was created by Claude.

544
01:04:48,880 --> 01:04:58,880
And because I don't program it in Python, so I said, hey, create a, a Python script for me that extracted the audio for, for the video.

545
01:04:58,880 --> 01:05:02,880
That's fine. We extracted the, the, the, the, the wave file.

546
01:05:02,880 --> 01:05:09,880
Now, then I have another file that say, hey, subscribe this wave file and create an SRT.

547
01:05:09,880 --> 01:05:30,880
And then I use AI just to translate this SRT to, to, to, to, on other languages. So instead of using a lot of token to do all the process, uh, using AI, I just use a small token just to, to, to, to translate the, the, the SRT files and the rest of the process I can run directing on, on, on my machine.

548
01:05:30,880 --> 01:05:44,880
I think, uh, having this, this mix, uh, it's very important because, uh, yeah, say, I can do everything, but doesn't mean you, you have to do everything using AI.

549
01:05:44,880 --> 01:05:54,880
Yeah, yeah, yeah, uh, I actually look for, uh, I like to build and voice depth tool for, for my YouTube videos.

550
01:05:54,880 --> 01:06:14,880
Yeah, that's fine. I have some good, good Python scripts with, uh, for example, uh, this podcast, we can, uh, cut all the, the silences from, from the video, we can analyze the, the, the audio layers and, uh, take off your, your language, uh,

551
01:06:14,880 --> 01:06:26,880
uh, accents, for example, everything like that. And this, this is a, uh, a, a systematic matter. This need to happen every time the same way.

552
01:06:26,880 --> 01:06:40,880
So this, for me, is programming. If you need to generate something new, if you'll need to have, uh, different approach from, from something, this is AI, this is the generative AI.

553
01:06:40,880 --> 01:06:48,880
Uh, some, yeah, we're a little bit over time, but I, uh, if it's okay, I have a rapid fire around. No, that's fine.

554
01:06:48,880 --> 01:06:52,880
Yeah. Um, single agent or multi agent.

555
01:06:52,880 --> 01:07:07,880
Uh, multi agents, uh, for, uh, I think for the most of complex job, uh, multi agents with, so be agents are, are the way to have, um, uh, prompt engineer context engineering.

556
01:07:07,880 --> 01:07:17,880
Uh, context engineering because it's way more complex and can deal with multiple, uh, stuff.

557
01:07:17,880 --> 01:07:22,880
Uh, loops are the future.

558
01:07:22,880 --> 01:07:28,880
Uh, in a way that, what is coming up for, yeah.

559
01:07:28,880 --> 01:07:42,880
Uh, you may know what's coming up from, from AI. Uh, I think we, we have more powerful agents, uh, more powerful models.

560
01:07:42,880 --> 01:07:53,880
And, uh, writing better code, writing better designs, for example, which one, something that I use in real, continue to evolve.

561
01:07:53,880 --> 01:08:09,880
We are, uh, in the beginning of the generative AI. So we can expect much more if you're impressive with AI nowadays, uh, in the next few years, you get more, uh, double time more impressive.

562
01:08:09,880 --> 01:08:18,880
Yeah, uh, under, if satan and dahlia call you and say you get all the money and resources to make dot net better, what feature will you develop?

563
01:08:18,880 --> 01:08:31,880
Uh, feature, uh, I don't know because, uh, the dot net is, uh, is a very good framework in my opinion that the team is, is doing a very good job.

564
01:08:31,880 --> 01:08:39,880
And we are increasing performance, increasing security. It's a very major, uh, for a framework.

565
01:08:39,880 --> 01:08:54,880
Uh, we don't have, uh, if you look to other frameworks, you'll see a lot of issues with security, for example, in that fact, we basically, doesn't have this, this kind of, uh, of things.

566
01:08:54,880 --> 01:09:00,880
So, uh, security and speed is, is always good.

567
01:09:00,880 --> 01:09:13,880
So I will invest in marketing to, to show the world how good is dot net. So I will pick a lot of money and say, hey, I want to, I want the people to know about that.

568
01:09:13,880 --> 01:09:27,880
And, uh, know how good, uh, stuff can be done in dot net. So let's reflect on some products that we have, uh, right, rewrite it in dot net using blazer using all the technology that we have.

569
01:09:27,880 --> 01:09:32,880
And, uh, show the word how good it is.

570
01:09:32,880 --> 01:09:42,880
Okay, and then my last rapid-fired route question is, uh, if I showed invite, uh, who, who shall I invite as next guest and what, uh, question should I ask?

571
01:09:42,880 --> 01:09:45,880
Wow.

572
01:09:45,880 --> 01:09:51,880
Uh, programmer or not programming for business?

573
01:09:51,880 --> 01:10:09,880
Uh, well, I'm very excited to, to, uh, programming. And, uh, one person that I, I really into the, the videos, the other is Mac, it just, uh, he created some good skills.

574
01:10:09,880 --> 01:10:23,880
And, uh, a, uh, an AI engineering with high level skills, uh, he talked, a lot of, uh, a, AI. So this is the, the, the person that I most watch the videos, not a.

575
01:10:23,880 --> 01:10:24,880
Yeah. Then, uh, yeah.

576
01:10:24,880 --> 01:10:24,880
And then, uh, yeah.

577
01:10:24,880 --> 01:10:26,880
And then, uh, thank you for, for joining me.

578
01:10:26,880 --> 01:10:38,880
Um, I'm talking, um, taking us from traditional software, development, all the way into AI agents, Microsoft, agent, product, world dot net, and the future of, uh, and gender software.

579
01:10:38,880 --> 01:10:46,880
Uh, yeah. One thing is, is, from years clear, AI isn't simply another library developer, um, add to an application.

580
01:10:46,880 --> 01:11:00,880
We are moving from where developers explicitly define every step forward, we are developed increased, we define goals, tools, context, emissions, and boundaries.

581
01:11:00,880 --> 01:11:11,880
AI deterministic, how, yeah, to accomplish the past, um, that change how we build software, how we test software, how we secure it and the potential.

582
01:11:11,880 --> 01:11:24,880
The, uh, why being a software development means for, for the first place. So, yeah, um, I say to all the guests, if you enjoy the conversion, follow the MC65 podcast, you find all the info, uh, under it in show notes.

583
01:11:24,880 --> 01:11:30,880
And, yeah, um, thank you for, for, for being here when you're one hour.

584
01:11:30,880 --> 01:11:33,880
Thank you very much, Verica was a pleasure for me.

585
01:11:33,880 --> 01:11:36,880
And congratulations for your.