From Demo to Production- Building Enterprise AI Agents That Actually Work with Microsoft Copilot Studio with Elliot Margot [MVP]

1
00:00:00,000 --> 00:00:07,920
Welcome back to the M665 podcast where we explore the world of Microsoft technologies,

2
00:00:07,920 --> 00:00:14,160
talking to people, they building a future of enterprise at e. Today's guest is Elliot Margot,

3
00:00:14,160 --> 00:00:19,360
Microsoft MVP for Microsoft C65, co-pilot and co-pilot studio,

4
00:00:20,240 --> 00:00:30,160
team lead for Jamstart, co-pilot agents at Vijo and one of the experts helping enterprise

5
00:00:30,160 --> 00:00:37,520
is moved to AI from existing demonstration to productive, ready business solution. We'll explore

6
00:00:37,520 --> 00:00:43,680
multi-agent architecture, Microsoft co-pilot studio, enterprise rack, MCP integrations,

7
00:00:43,680 --> 00:00:51,680
governance security and so on and so on. We hope we get all done. Yeah, if you're building

8
00:00:51,680 --> 00:00:57,680
Microsoft co-pilot solutions or wondering how to move beyond simple chatbots into

9
00:00:57,680 --> 00:01:03,680
intelligence enterprise agents, this episode is for you. So let's start. Welcome Elliot.

10
00:01:03,680 --> 00:01:07,120
Thank you, great intro, love the first time.

11
00:01:09,280 --> 00:01:16,320
Yeah, can you tell us a little bit about yourself and your journey into the Microsoft technology?

12
00:01:16,320 --> 00:01:23,280
I started about Microsoft. I started very on using Microsoft and everything.

13
00:01:23,280 --> 00:01:30,560
As most people have a first computer was also running on Windows and really start working

14
00:01:30,560 --> 00:01:36,960
with co-pilot and everything two and a half years ago. Before I got into generative AI, I was

15
00:01:36,960 --> 00:01:43,280
already a product owner for insurance chatbots and I really wanted to move into GNII and start

16
00:01:43,280 --> 00:01:50,800
implementing and really build solutions. So I switched job and really started doing that.

17
00:01:50,800 --> 00:01:58,640
So I started at 27 as a junior consultant. I was a bit of a rough transition but it was a lot of fun

18
00:01:58,640 --> 00:02:05,040
and a big steep learning curve and I like learning, I like learning fast.

19
00:02:06,320 --> 00:02:14,480
So I put a little bit of my own into it and two years later, here I am a year ago, I started

20
00:02:14,480 --> 00:02:21,440
leading the team for all that is integration governance, use case, hackathon, workshops,

21
00:02:21,440 --> 00:02:26,160
agent with them as well and famous prompt sounds now that they have so many different names for it.

22
00:02:26,160 --> 00:02:30,960
So many different variants but yeah, so that's my little journey through it.

23
00:02:31,680 --> 00:02:41,920
Yeah, you working in your company are, I don't know, you leading the Microsoft Jumpstart program.

24
00:02:41,920 --> 00:02:44,800
What exactly is Jumpstart?

25
00:02:44,800 --> 00:02:53,120
So Jumpstart is pretty wide. It's anything to do, well essentially it's Microsoft

26
00:02:53,120 --> 00:02:57,840
official acceleration program. It changed this year, it's called Frontier Accelerator.

27
00:02:58,320 --> 00:03:04,880
It's fully funded by Microsoft for eligible enterprise customers. So engagement is organized

28
00:03:04,880 --> 00:03:15,120
in essentially five six places. So you inspire co-pilot use to collaborators. You assess the level

29
00:03:15,120 --> 00:03:20,560
that's currently in the company, you design solution, then you build them, you do workshops,

30
00:03:20,560 --> 00:03:28,000
you do rollouts. No, it's not so much ACM which is nice. Don't like acting as a level one

31
00:03:28,000 --> 00:03:37,280
service desk but essentially it's, this part is great. And finally you have a little bit of everything.

32
00:03:37,280 --> 00:03:44,960
The team operates with AI native tooling, MCPs and everything that help us be on top

33
00:03:44,960 --> 00:03:52,400
and we try to inspire two demos and really bring co-pilot to organizations.

34
00:03:53,120 --> 00:03:59,040
Yeah, sometimes I have a little bit problems with the naming convention by Microsoft.

35
00:03:59,040 --> 00:04:01,280
Very things to name co-pilot.

36
00:04:01,280 --> 00:04:15,120
Sometimes it's a little bit hard to have an overview. But how did you, especially in co-pilot,

37
00:04:15,120 --> 00:04:25,920
staff, how did you not get over-vimed with all these updates? I did. There's a lot.

38
00:04:25,920 --> 00:04:36,000
But there are ways to co-pilot to find information and so on. I built my own little workflow which

39
00:04:36,000 --> 00:04:44,080
sends every day at nine, a little information from anything from, and I have to get it back because

40
00:04:44,080 --> 00:04:50,240
it's so big. So it goes from anything from Microsoft co-pilot studio to M365 co-pilot and apps,

41
00:04:50,240 --> 00:04:58,080
agents, co-work, front-to-air program, extensibility, SDKs, work IQ, governance, foundry, power

42
00:04:58,080 --> 00:05:06,800
platform, and then the entire M365 suite. So like teams, outlook, office apps, fever, loop, and etc.

43
00:05:06,800 --> 00:05:13,360
So I would love to share my screen but I'll show you the workflow but essentially every day.

44
00:05:13,840 --> 00:05:21,520
That's all you only about. I can invite you to the M365 live stream podcast. So

45
00:05:21,520 --> 00:05:31,600
then we can look a little bit more into it. But you work at, I don't know, whitey-bow?

46
00:05:31,600 --> 00:05:38,880
Yeah, which, which of you? It comes from the English word width, then width, if you,

47
00:05:38,880 --> 00:05:45,680
and then the name, the main name which was all the craze was .io. So it was widthsiv.io and then

48
00:05:45,680 --> 00:05:56,320
it changed to VTV. Okay. And it's called AI Solutions for Microsoft 65. The A1 suite for S365.

49
00:05:56,320 --> 00:06:06,320
What exactly did you do or what did the company do? So we've got, it's quite large. We mostly do

50
00:06:06,320 --> 00:06:14,160
service now, which we also have products. So like products for teams, teams app. We also went into

51
00:06:14,160 --> 00:06:21,200
the custom co-pilot. So we have got a few co-pilots. Like my first demo was Bot Murmades and I was

52
00:06:21,200 --> 00:06:25,440
like, I told VTV that that would be great if we could have that in the Asian store and sell it to

53
00:06:25,440 --> 00:06:31,440
customers. So like a co-pilot agent, which only does diagrams and stuff. So we do have on-shelf

54
00:06:31,440 --> 00:06:38,880
products, which we do mostly service custom stuff for, for big companies. Yeah, your company sits

55
00:06:38,880 --> 00:06:46,480
in France and you're living in the Spitzerland. Yes. So we got some most of the team in France and we

56
00:06:46,480 --> 00:06:54,320
got so few people in Switzerland, including the CEO. So it's, it's being a little bit international

57
00:06:54,320 --> 00:07:09,920
of a country. European. Yeah, safe Texas. Joke, joke. So, yeah, there is a topic and a lot of enterprises

58
00:07:09,920 --> 00:07:22,880
use AI, I say, yeah, they try to use AI. Why is especially in enterprises, the users of AI,

59
00:07:22,880 --> 00:07:33,040
so I say, yeah, hard from your perspective. Because they want to be able to know what goes on with

60
00:07:33,040 --> 00:07:38,720
the agents and that's the governance part. And governance is very hard because it's very big.

61
00:07:38,720 --> 00:07:45,200
Every different agent you want to implement has a different way to track. You want to be able to

62
00:07:45,200 --> 00:07:51,200
find the thing that you can implement and not touch for the next two years, which is,

63
00:07:51,200 --> 00:07:56,720
unfortunately, not very realistic because it evolves so quickly. So it's a full-time job for people,

64
00:07:56,720 --> 00:08:02,160
for administrators, which system admins, which have to keep up and everything and it's not even their

65
00:08:02,160 --> 00:08:07,440
primary job most of the time. So it's very much always staying on top, being able to follow everything

66
00:08:07,440 --> 00:08:14,480
that goes on, knowing that nothing is leaked, that the DLP data loss protection prevention policies

67
00:08:15,040 --> 00:08:23,520
are solid and in place and that's the hardest thing to put in place. Yeah, I love the governance

68
00:08:23,520 --> 00:08:31,040
topic about that. I think a little bit about, when we talk about AI,

69
00:08:31,040 --> 00:08:44,080
I try to figure out what I mean, but I think the best question for this is

70
00:08:44,560 --> 00:08:58,720
why should companies focus beef on governance and compliance and security before they start?

71
00:08:58,720 --> 00:09:08,640
It's a bit, I want to say, ironic. You can't see the results of the governance

72
00:09:08,640 --> 00:09:18,160
in place if you don't start. So you can have to start while building it. The best, the companies who

73
00:09:18,160 --> 00:09:23,440
learn the quickest are companies who do it themselves and are in scared to, I want to say, get

74
00:09:23,440 --> 00:09:32,320
wet a little bit in the pool because it's, as I said, you can't really get information out of it

75
00:09:32,320 --> 00:09:43,200
without having data. So governance is important to set up, but if you focus too much on it at the start,

76
00:09:43,200 --> 00:09:53,840
you're losing time that you can learn on the job kind of and it's a shame because you can't really

77
00:09:53,840 --> 00:10:04,240
learn something without doing it. Yeah, I think that is, yeah, I think a lot of companies doing it,

78
00:10:04,240 --> 00:10:13,440
they say, I know it from my companies, I work for, they say, "Oh, then AI, we don't use it because

79
00:10:13,440 --> 00:10:22,320
it's not secure and so on." And the other side, I have, we activated for all the people,

80
00:10:22,960 --> 00:10:33,680
and don't give a fuck on something like the app's enabling. This is really, really funny, I think.

81
00:10:33,680 --> 00:10:47,280
So I think a little bit about, when I go to, as it comes out, I go, I don't know why the

82
00:10:48,400 --> 00:10:55,200
yeah, oh, no, I changed, we have two parts, we have the guys they like to use AI and they have

83
00:10:55,200 --> 00:11:04,960
the guys, "Oh, would it kill my job? I don't like to use it." And I say, or what, how did you,

84
00:11:04,960 --> 00:11:13,760
this is cool, how did you sell the AI to their executive sponsors?

85
00:11:18,080 --> 00:11:23,440
It's not easy because you do have the, you've got two facets, you've got the first facets, which is like,

86
00:11:23,440 --> 00:11:30,400
it's going to have a good ROI, it's going to improve efficiency, but it's not going to steal jobs.

87
00:11:30,400 --> 00:11:35,600
If you want something that's going to do the job, somebody, you're not looking at a salary, you're

88
00:11:35,600 --> 00:11:41,120
looking at a little bit more than a salary, because it needs to run the entire time, it's going to,

89
00:11:41,120 --> 00:11:47,120
to cost you a lot, and even nowadays with the newer models and like the Chinese models came in,

90
00:11:47,120 --> 00:11:51,200
which you can host locally, it still takes a lot of electricity to run.

91
00:11:51,200 --> 00:12:02,320
So, the biggest question isn't so much the, the ROI at some point, it's kind of the, the wow effect,

92
00:12:02,320 --> 00:12:08,480
and also the, how much time is going to save me? It's a little bit of the ROI, but it's a lot of the

93
00:12:08,480 --> 00:12:16,000
wow effect, and that's kind of the, the angle that I push. So let's take, for example, the IT help desk

94
00:12:16,000 --> 00:12:22,880
agents, you know, you've got like 50 people, or maybe not 50 people, you've got like five people

95
00:12:22,880 --> 00:12:36,400
working ITL desk level one, they take like 13,000 tickets a year maybe, and let me actually do some

96
00:12:36,400 --> 00:12:44,240
math on the side, so I don't go something completely crazy. So you've got 10 people and these 10 people

97
00:12:44,240 --> 00:12:55,280
do 15,000 tickets. So that's 1,500 tickets per, per user per, per year. So that's quite a lot,

98
00:12:55,280 --> 00:13:04,800
a lot of tickets. So if your goal is to say, okay, let's produce that by 10%, so one ticket takes you,

99
00:13:04,800 --> 00:13:13,600
let's say, twin, let's be realistic a little bit, it takes you 30 minutes, right? So if 30 minutes,

100
00:13:14,240 --> 00:13:20,400
is still about like average wage, let's say like $20. So $20 for a ticket, if you reduce

101
00:13:20,400 --> 00:13:30,640
just 15,000 tickets, that's 20 times, don't need, don't need a calculator for that, that's 30,000 per year.

102
00:13:30,640 --> 00:13:37,600
So you already have like a little, little good thing, and then it really depends what kind of model

103
00:13:37,600 --> 00:13:44,240
you want to use behind, is it, are you okay with winning money, are you okay with optimizing the workforce,

104
00:13:44,240 --> 00:13:51,520
so you can take a bigger model, which can give a better quality. And the goal is kind of also to

105
00:13:51,520 --> 00:13:57,040
make sure that they understand everything that goes with the executives, understand everything in

106
00:13:57,040 --> 00:14:05,840
the back, transparency and trust. Yeah, I find this really, really interesting, Carlos, I use the

107
00:14:05,840 --> 00:14:18,240
Microsoft Plurity on my page, and I see I have nearly 30,100 and so on, AI citations, and that means

108
00:14:18,240 --> 00:14:28,240
every minute, every minute, it's answered 21 questions, so this is amazing. I don't know, then there's only

109
00:14:28,240 --> 00:14:36,400
one minute when we see, or when we think you can answer a question from a client in one minute,

110
00:14:36,400 --> 00:14:53,200
you need 21 dudes, did you think we losing jobs with AI, or is it, you get more efficient?

111
00:14:55,280 --> 00:15:02,080
I don't think we're going to lose jobs with AI, I think the kind of, as my own opinion,

112
00:15:02,080 --> 00:15:07,120
but I think that we're going to replace the workforce with AI was kind of the sales pitch, it wasn't the

113
00:15:07,120 --> 00:15:14,960
goal. AI is, as we've seen the trend, it's always getting more and more expensive, it's not getting

114
00:15:14,960 --> 00:15:19,600
cheaper at some point, because you still need the hardware to run it, the electricity, the infrastructure,

115
00:15:19,600 --> 00:15:27,360
the water, and it's going to make people have more time for things, but it's still going to be

116
00:15:27,360 --> 00:15:33,920
expensive, so you need to find the right way to explain that as well, and I think that's also

117
00:15:33,920 --> 00:15:43,760
where it's very interesting. Yeah, I think sometimes what we not discuss is with AI,

118
00:15:45,040 --> 00:15:52,480
AI can still lose external knowledge, but they don't have the internal knowledge about processing,

119
00:15:52,480 --> 00:16:00,480
about, I don't know, about the code from a SaaS company, or how the, I don't know, the

120
00:16:00,480 --> 00:16:09,440
refrigerator or the door works, so I think, yeah, people become more efficient,

121
00:16:11,520 --> 00:16:19,200
and there, I think there are a lot of people actually outside and say, oh, I run my own company,

122
00:16:19,200 --> 00:16:29,600
will, will the agent staff, so why, why does multi agent systems becoming one of the hottest

123
00:16:29,600 --> 00:16:39,440
topic in AI, what did you think? So let's do a little example of an agent that builds a powerpoint

124
00:16:39,440 --> 00:16:46,400
slide, a powerpoint deck. There's going to be many errors in that, and the first shot, you need to

125
00:16:46,400 --> 00:16:51,520
go back in, you need to improve it, it's going to take tokens, it's going to take your time, and

126
00:16:51,520 --> 00:16:57,040
you do it once, it takes a bunch of time, you do it twice, it takes last time, you do it a hundred times,

127
00:16:57,040 --> 00:16:59,520
it takes almost zero because you automate the process.

128
00:16:59,520 --> 00:17:08,160
What if you can have one agent that creates a slide, another that audits the slide,

129
00:17:09,120 --> 00:17:13,680
so you have marketing executive, you have a UX designer, you have a developer, you have a product

130
00:17:13,680 --> 00:17:19,760
manager, all of these different personalities that look at the slide, give feedback, and it ought to

131
00:17:19,760 --> 00:17:25,520
improve. And that's kind of where the multi agent system comes into play in my opinion, it's kind of

132
00:17:25,520 --> 00:17:32,320
the self-healing, self-improven system, and I think that's a little bit where we're heading.

133
00:17:33,040 --> 00:17:39,760
Yeah, that's interesting, so my question is then, why we have all the specialized agents of

134
00:17:39,760 --> 00:17:48,560
dome build, big universal item or brain that's, yeah, no, no, it's all.

135
00:17:48,560 --> 00:17:53,920
Because it's easier to sell.

136
00:17:57,520 --> 00:18:04,720
I mean, if you're a company that makes only agents for medical, you're going to have to go to doctors,

137
00:18:04,720 --> 00:18:12,880
go to hospitals and say, I've got this, I've learned as many hospitals as there are companies,

138
00:18:12,880 --> 00:18:19,840
but you make generalist doctors, you'll be able to sell it as a first, maybe input, you make

139
00:18:19,840 --> 00:18:30,080
cardiologist, it's going to be very specific, same thing for general LLMs, instead of like specific

140
00:18:30,080 --> 00:18:36,800
pre-trained fine-tuned and stuff like that, there's the market, there's just less of a market,

141
00:18:36,800 --> 00:18:42,880
so they have to be more expensive, it's more expensive to train, it takes time, it takes efforts,

142
00:18:43,520 --> 00:18:52,080
it maybe some also take more compute, and I think that's why general all knowing, well, they say

143
00:18:52,080 --> 00:18:57,280
they're all knowing, at least, till they hallucinate, kind of sells better.

144
00:18:57,280 --> 00:19:06,400
Yeah, yeah, I have, use the medical, I have a really cool example, there was a company,

145
00:19:06,400 --> 00:19:19,840
Ambus, they started as a company that helps people in doctors, nurses, and so on, and now they have

146
00:19:19,840 --> 00:19:29,360
all the studies and find, yeah, anomalies like, okay, in the studies, all the people have,

147
00:19:29,360 --> 00:19:35,920
I don't know, it acts, and they're the black pressure, they're going down, when people, this is really

148
00:19:35,920 --> 00:19:52,000
cool, I think, yeah, so, but, I think a little bit about it, and one thing to do, this is really

149
00:19:52,000 --> 00:19:59,040
cool, it's orchestration, what orchestration patterns have proven most successful from your

150
00:19:59,040 --> 00:20:14,240
perspective? It really depends, but I think the biggest factor is being able to communicate,

151
00:20:14,240 --> 00:20:19,440
well, what you want, being able to describe exactly what you want, the workflow you want to design,

152
00:20:19,440 --> 00:20:27,200
and being a good, good communication person, that was not good communication,

153
00:20:27,200 --> 00:20:34,800
but being able to express yourself in direct and straight manner without being ambit usable,

154
00:20:34,800 --> 00:20:40,560
what you want, that's the best orchestration that you can have, it doesn't matter if you use

155
00:20:40,560 --> 00:20:47,520
crew or a line chain or something else, as long as you don't describe what you want, what you want

156
00:20:47,520 --> 00:20:55,600
to do, the tasks of every single agent in your workflow is useless, you go to the top right and you press

157
00:20:55,600 --> 00:21:07,200
bin, yeah, when we talk about AI, we all talk about large language, how good they are, how good they

158
00:21:07,200 --> 00:21:15,360
are, compare it, and so on, but there are some basics, I think, machine learning, and one topic,

159
00:21:15,360 --> 00:21:29,760
it's a rock, so how will you explain rock, and how did you see the development of rock, especially

160
00:21:29,760 --> 00:21:41,120
when we talk about agentic rock? That's a very wide topic, but you can think of rock as,

161
00:21:43,520 --> 00:21:48,400
you've got a library of books, each book has a title, it has a specific color,

162
00:21:48,400 --> 00:21:55,920
and then you start adding what we call meta tags, meta information, so you start adding a tag,

163
00:21:55,920 --> 00:22:02,800
like science fiction, themes, the stuff like that, and you start building a little bit of an

164
00:22:02,800 --> 00:22:08,000
information base about what it's about, so that way the agent when you ask,

165
00:22:11,760 --> 00:22:18,880
where is the, well, a question about HR, like when is my next holiday? It knows that it needs to go

166
00:22:18,880 --> 00:22:24,960
get the book about HR, and kind of that one it doesn't go into operations or something like that,

167
00:22:24,960 --> 00:22:31,840
and that's probably the easiest way to describe it, but the evolution of rock has,

168
00:22:31,840 --> 00:22:38,640
in my opinion, not done leaps, but it improved and everything, but still

169
00:22:40,320 --> 00:22:47,760
you go find the H1, you go find the header 2, header 3, you look for keywords in something,

170
00:22:47,760 --> 00:22:54,480
maybe there's some indexing that is done before, notably like with the metadata and stuff like that,

171
00:22:54,480 --> 00:22:58,880
if you go to my website and talk to my agent, it's able to like guide you through the pages and

172
00:22:58,880 --> 00:23:07,360
everything, because I pre-did the indexing and stuff like that, but I'm not sure I can answer your

173
00:23:07,360 --> 00:23:15,520
question quickly and still have time for everything else. Yeah, this was cool overview, but what did you

174
00:23:15,520 --> 00:23:23,120
think, how important are metadata and how handled you ride? I didn't understand the answer.

175
00:23:23,120 --> 00:23:27,440
How important is metadata and how to handle it's ride?

176
00:23:27,440 --> 00:23:36,320
metadata is key, because if you don't, let's take, let's go into co-piles to you.

177
00:23:36,320 --> 00:23:39,520
If you take a co-piles studio and add SharePoint based, knowledge based,

178
00:23:39,520 --> 00:23:47,440
you add PDFs which are full text, doesn't have any headers, doesn't have any information,

179
00:23:47,440 --> 00:23:53,680
if you go into the settings and play around a little bit, you can see that there are specific settings

180
00:23:53,680 --> 00:23:58,480
that you can add, select its index better. So for example, you can add a new column and put

181
00:23:58,480 --> 00:24:04,240
description, you can add a new column and put language, and with those parameters, it's able to,

182
00:24:04,960 --> 00:24:10,320
if somebody speaks with it in Chinese, simplify Chinese, it's able to go to the right document that's

183
00:24:10,320 --> 00:24:16,960
already in Chinese, it doesn't have to translate it and re-translate it behind. So it's important, not only for

184
00:24:16,960 --> 00:24:24,400
quick retrieval, but also token optimizations, so optimizing your costs and everything long-term.

185
00:24:24,400 --> 00:24:30,640
So good data, good output, bad data, bad output, through STPG.

186
00:24:31,200 --> 00:24:40,240
Yeah, it's, yeah, it's a particular regulator for a bad data, I think, it works very well.

187
00:24:40,240 --> 00:24:54,000
So I think about, you know, an X-L version, 13, Vellarin, I don't know, X-approved,

188
00:24:56,800 --> 00:25:02,800
how can we handle a versioning effectively?

189
00:25:02,800 --> 00:25:11,280
The most simple thing is you crushed the old version.

190
00:25:11,280 --> 00:25:20,240
But if you want to have a history and everything, that's not the right place to put everything,

191
00:25:20,240 --> 00:25:25,360
if you put everything into a rag place where you're going to do the rag, it needs to be

192
00:25:25,840 --> 00:25:31,440
the thing that's you're ready to go to production. So if you've got a V19 and a V12,

193
00:25:31,440 --> 00:25:38,880
sort of V19 and a V20 of the documents in your production rag environment where you can go

194
00:25:38,880 --> 00:25:44,640
to get information, you really need to have the V19 somewhere else and the V20 really in your

195
00:25:44,640 --> 00:25:52,560
production place. Again, it could get the 19 and put out the information or the 20 and you're

196
00:25:52,560 --> 00:25:57,840
kind of taking risk and also the agent is going to look at both and go, oh, this is newer and it's

197
00:25:57,840 --> 00:26:03,680
going to consume more tokens and your costs are going to go up again. So really good data management is key.

198
00:26:03,680 --> 00:26:13,680
Yeah, and I think when we talk about rag, what role do you chunking play?

199
00:26:19,920 --> 00:26:26,560
I want to say chunking is great, but it doesn't always get the right information depending on how your

200
00:26:26,560 --> 00:26:34,560
data is made. So if you take like 150 pdf, it's going, I don't know, different chunkings have

201
00:26:34,560 --> 00:26:40,720
different ways. Maybe it takes every third word, maybe it takes every second paragraph, and it misses

202
00:26:40,720 --> 00:26:47,600
like the key thing in the middle because it skipped something. Some real found of like having good

203
00:26:47,600 --> 00:26:54,000
metadata, tagging your information well, having a little description like in this pdf, you will find

204
00:26:54,000 --> 00:27:00,480
the HR policy about absences and sick leave, and then the other one says you will hear you will find

205
00:27:00,480 --> 00:27:06,880
absences and maternity leave. And that way you have keywords in your description, which can match

206
00:27:06,880 --> 00:27:13,840
easily and then it goes, it knows it goes here. But yet, I'm not a big big fan of chunking documents.

207
00:27:16,560 --> 00:27:24,800
Yeah, keywords. So one of the hardest keywords is actually MCP, so I have repeated from Google

208
00:27:24,800 --> 00:27:36,560
sometimes, MCP, MCP. Well, Joke, MCP comes one of the most discussed AI standard. What is MCP?

209
00:27:36,560 --> 00:27:47,680
What is MCP? So to put it simply, you've got multiple APIs, so you've got multiple endpoints you can call.

210
00:27:47,680 --> 00:27:57,200
For example, let's take your fridge. Let's go into reality. Your fridge has multiple levels.

211
00:27:57,200 --> 00:28:04,800
So you go level one, level two, level three, level four, and you can go into any of the levels and

212
00:28:04,800 --> 00:28:11,680
pick yogurt for Americans you can pick out eggs, who we keep them outside, maybe now with this heat,

213
00:28:11,680 --> 00:28:21,440
and so on. An MCP would enable an agent to instead of you plugging everything at once and having

214
00:28:21,440 --> 00:28:26,560
to scan every level at once, it goes, okay, I've got all the levels that are available,

215
00:28:26,560 --> 00:28:32,240
and I know exactly where to find things. I don't have to look again every time through everything.

216
00:28:34,240 --> 00:28:45,200
Yeah, and why is MCP so important for enterprise AI? Because it's something that is that for groups,

217
00:28:45,200 --> 00:28:51,280
multiple APIs, multiple levels of the fridge, and you can just plug it into something instead of

218
00:28:51,280 --> 00:28:58,160
having to plug every single level at one time. So it's just a bundle of tools you can call

219
00:28:59,120 --> 00:29:06,000
different refrigerator levels and really just have something that's secure, authentified instead of

220
00:29:06,000 --> 00:29:12,480
again, authentifying every single level. It's like having a camera in front of your fridge with

221
00:29:12,480 --> 00:29:18,320
facial recognition, so it will open for you instead of having one every level and going, can I take

222
00:29:18,320 --> 00:29:28,240
this yogurt out and goes, it's not you, but you already opened the fridge. So a lot of people don't

223
00:29:28,240 --> 00:29:36,880
understand the different tradition between our traditional APIs and MCP. What does the real

224
00:29:36,880 --> 00:29:46,480
define? So the real difference is traditional API, you don't necessarily have a description for it.

225
00:29:46,480 --> 00:29:56,080
So it doesn't know what synates when to call it, how to use it. If you were to make multiple APIs

226
00:29:56,080 --> 00:30:01,760
for writing emails, you would maybe have one to draft an email, another one to send an email. So you

227
00:30:01,760 --> 00:30:10,320
would have to explain how to use them. In the description, you can put, for example, use HTML to draft

228
00:30:10,320 --> 00:30:20,160
the email, don't use m-, don't forget the accents, maybe not very valid in English, but you're able to

229
00:30:20,160 --> 00:30:27,120
bundle your information well so that the agent knows how to use the tool effectively and really

230
00:30:27,120 --> 00:30:33,840
customize it a lot more than you could just do with one single endpoint. And again, it's bundled

231
00:30:33,840 --> 00:30:42,400
already, it's fixed, so it's easier to plug and play. I think one topic, it's not the sexiest in the

232
00:30:42,400 --> 00:30:49,360
world, but a lot of companies have to talk about governance, security and trust in AI systems.

233
00:30:49,360 --> 00:30:59,360
So when we start with governance, what governance showed organization established before they

234
00:30:59,360 --> 00:31:16,720
employing AI agents? Microsoft agents? Yeah, Asian 365 license. But seriously, without that license,

235
00:31:16,720 --> 00:31:20,080
you'd like to lose security information and stuff like that, which is critical.

236
00:31:20,080 --> 00:31:28,640
Being able to take an hour before you start deploying agents and familiarize yourself with

237
00:31:28,640 --> 00:31:37,840
admin 365, the co-pilot interface, where your data goes, purview, clarity, all that kind of ecosystem,

238
00:31:37,840 --> 00:31:44,720
at least being familiar with it, knowing which look, information, preferably connecting your

239
00:31:44,720 --> 00:31:50,480
favorite AI agent to Microsoft and CPLern. So that's so you can ask it quickly, where do I find this

240
00:31:50,480 --> 00:31:55,760
information and goes, find it and goes, you can find it here. And that will save you a lot of time

241
00:31:55,760 --> 00:32:03,520
in the long run, so kind of preparing yourself mentally and preparing the terrain for everything

242
00:32:03,520 --> 00:32:13,120
that comes afterwards. Yeah, yeah, yeah. Okay, we are on the bad topics, so let's talk about the most,

243
00:32:13,840 --> 00:32:21,600
at the most tightest topic in data loss prevention, VLT. How can we handle this, the co-pilot's to you?

244
00:32:21,600 --> 00:32:27,920
So it's power platform, the LPs that you set up for the environment itself.

245
00:32:27,920 --> 00:32:34,400
So for example, what I recommend is if you're going to have a co-pilot agent that goes outside of

246
00:32:34,400 --> 00:32:42,400
Microsoft, so for example, those HTTP requests, you pass through an app, which in that app only is

247
00:32:42,400 --> 00:32:52,960
allowed to use this or that. And it can't decide to go query MSN weather or like try to send many

248
00:32:52,960 --> 00:33:02,800
payloads to your bank, but something that could be dangerous. Maybe you want to remove a MailChimp,

249
00:33:02,800 --> 00:33:11,840
if it's not in your scope, really keep the scope small, so that you're not able to use things that

250
00:33:12,400 --> 00:33:20,240
you shouldn't be able to. So if your use case is an agent that helps your productivity by summarizing

251
00:33:20,240 --> 00:33:26,640
your day from teams, your meetings, your emails, then you only leave teams,

252
00:33:26,640 --> 00:33:37,120
outlook. And I forgot the third one, but the point is there, you really put a small scope to it,

253
00:33:38,720 --> 00:33:42,800
so that it can't be out. And if you've got another use case, you create new environments,

254
00:33:42,800 --> 00:33:52,480
and you create new DLP for it. Let me drop a tool in the conversation, Microsoft Viewer.

255
00:33:52,480 --> 00:34:00,320
Absolutely, that too. But that's more global DLP than just co-pile Studio DLP,

256
00:34:00,320 --> 00:34:07,360
which is some power platform, but absolutely, PerView is key, making sure that you're just really dumb,

257
00:34:07,360 --> 00:34:12,720
but keeping your sensitivity labels on documents. If you've got something that's classified secret

258
00:34:12,720 --> 00:34:21,120
or confidential, maybe you don't want your agents to have access, like HR docs, personal HR documents,

259
00:34:21,120 --> 00:34:32,000
have them well classified and stuff like that. Yeah, yeah. Today, I have a really good day, because

260
00:34:34,320 --> 00:34:41,680
I have guests before. He built the master over, and the other topic I really, really love is

261
00:34:41,680 --> 00:34:50,160
it's the EGR you work for. Can you or doesn't ring a bell?

262
00:34:50,160 --> 00:35:00,400
Or it's ITER? I don't know how pronounced it right.

263
00:35:02,400 --> 00:35:08,720
Do you mean Shell? The company? Yeah. Yeah. Okay, no,

264
00:35:08,720 --> 00:35:15,120
no, I haven't worked for them yet. Okay, okay, then I was wrong, wrong, wrong, for my,

265
00:35:15,120 --> 00:35:24,400
yeah, for my research, the server for this. But you also have a multi-tenant agent for generality,

266
00:35:24,400 --> 00:35:34,080
did I hope you're right? Yeah, so multi-tenant agents are kind of fun, especially for big holding

267
00:35:34,080 --> 00:35:40,160
companies where they have like maybe like a holding company and then multiple little little

268
00:35:40,160 --> 00:35:48,800
subsidiaries. And yeah, no, those are those are absolutely fun, especially when it comes to finance

269
00:35:48,800 --> 00:35:54,640
and stuff like that, where they have to report everything up and you have to make sure that the agent

270
00:35:54,640 --> 00:36:00,560
understands everything and is able to enter the right information or update the right information.

271
00:36:00,560 --> 00:36:05,520
And when you've got one holding company with like 150 subsidiaries,

272
00:36:05,520 --> 00:36:15,680
it gets very much quick and intense very quickly. Yeah, Microsoft has this power platform and

273
00:36:15,680 --> 00:36:22,960
everyone is now a developer. We call the citizens of the developer. Yeah.

274
00:36:22,960 --> 00:36:32,560
How would it you think the citizen development has changed our development and especially

275
00:36:32,560 --> 00:36:40,720
AI development? I think that there are two things, a little bit back, you were able to ask,

276
00:36:40,720 --> 00:36:45,680
20p2 or you LLM, how do I build this? Help me build this and you will tell you, oh, yeah,

277
00:36:45,680 --> 00:36:51,840
first you have your trigger, you go when an email arrives, resume the email with like a coparist to

278
00:36:51,840 --> 00:36:56,400
do that you invoke and then send me a message and team saying, you got an email, this is what it's

279
00:36:56,400 --> 00:37:06,160
about, it's not urgent or it's urgent. Three steps, very quick. Now it goes a little bit further,

280
00:37:06,960 --> 00:37:13,920
where people aren't really building their own flows anymore, where they log in with Azure

281
00:37:13,920 --> 00:37:25,760
Key, CLI and you're able to build the flows with Cod or with VS Code with GitHub Copilot and it goes

282
00:37:25,760 --> 00:37:31,680
and plans, you want to do this? So I understand the scope and I'm going to build it for you and it goes

283
00:37:31,680 --> 00:37:36,000
into your tenants, your environment where you're accessed, where you're the owner of it. It's like a

284
00:37:36,000 --> 00:37:42,240
personal developer environment and it's able to say, okay, I'm going to build it for you, it builds it

285
00:37:42,240 --> 00:37:50,240
and most of the time you need to re-log correct a few things, add connections, but it gets much

286
00:37:50,240 --> 00:37:58,320
easier to build something on literally no code solution because you're not really building it anymore,

287
00:37:58,320 --> 00:38:04,880
you're just saying what you want to build and it's like vibe coding but kind of more

288
00:38:05,840 --> 00:38:16,640
vibe building if you get my drift. It's like, yeah, I think I don't like the word, but the

289
00:38:16,640 --> 00:38:23,920
prompt engineering was one of the other houses job for, I don't know, three years and now

290
00:38:23,920 --> 00:38:33,600
everyone should do it, but what from your perspective, what makes, I say, why code it?

291
00:38:33,600 --> 00:38:43,120
AI app, how did you check if it's a manufacturer, you have your checklist or what do you do?

292
00:38:43,120 --> 00:38:50,480
No, it really depends on the use case on what it's supposed to do, but automating the tests.

293
00:38:50,480 --> 00:38:55,440
So you've got like, for example, a big file where you just send a bunch of prompts and look at the

294
00:38:55,440 --> 00:39:02,960
outputs and also, and I guess I can't stress this enough, but manually testing the things you build

295
00:39:02,960 --> 00:39:11,520
is key to building something that is stable. If you just trust AI to monitor AI without you ever

296
00:39:11,520 --> 00:39:19,840
checking, that is how, like for example, the latest GPD model gets out of their sandbox and starts

297
00:39:19,840 --> 00:39:26,080
going to hug and face and hug and their face and stuff like that. It's very much important that

298
00:39:26,080 --> 00:39:32,400
you check what goes on with what you build. It's not, oh, I built it, that GPD or cloud or

299
00:39:32,400 --> 00:39:38,000
or get up copilot, so agent tells me it's fine, it runs, it's going to work in production, I trust it,

300
00:39:38,000 --> 00:39:43,520
I'm not going to do anything. Testing manually is very important, so that you can say yes,

301
00:39:43,520 --> 00:39:49,600
it's going to work and when you say yes, you are 100% sure that it will work.

302
00:39:51,360 --> 00:40:03,760
Awesome, awesome. There's one thing I see, it's the RFP agent. Can you tell a little bit about this?

303
00:40:03,760 --> 00:40:09,520
Absolutely, so RFP agent is an agent that automates purchases from 2 to 50k,

304
00:40:09,520 --> 00:40:17,680
so the user says what they want to buy, it redacts the RFP proposal, the documentation for it,

305
00:40:18,400 --> 00:40:27,440
and it sends the emails. After that, the suppliers respond, you receive the offers, the questions and so on,

306
00:40:27,440 --> 00:40:35,440
and you get everything analyzed with the GPD model, different GPD models specialized for different

307
00:40:35,440 --> 00:40:45,840
things. The user just has to look at what comes in, the offers, the questions, response to the questions,

308
00:40:46,560 --> 00:40:50,080
of course the question is already, here's the proposed response by the AI,

309
00:40:50,080 --> 00:40:59,600
and you just modify it a little bit, press send, sends the response, and until the deadline of the

310
00:40:59,600 --> 00:41:07,920
RFP, on the deadline you select the supplier winner, if you want to organize meetings to like

311
00:41:07,920 --> 00:41:14,800
discuss the offer as you can, and then you select it, emails go out, you were selected, you were not

312
00:41:14,800 --> 00:41:21,120
selected, and that's it. Yeah, over and I think it's really cool you become with the RFP agent

313
00:41:21,120 --> 00:41:26,880
and of the official Microsoft customer success story. Yes, that was my first project actually.

314
00:41:26,880 --> 00:41:35,200
So here was a big first project. Yeah, yeah, yeah. What have you learned on the side, have you

315
00:41:35,200 --> 00:41:45,440
had done this project? Power Automate can there be very tricky. That's the first thing I learned

316
00:41:45,440 --> 00:41:55,760
from loops to variables to compose actions to data manipulation, at the start I would use AI

317
00:41:55,760 --> 00:42:03,840
builder to kind of help me manipulate data, and then I got really into Power FX, and for those who

318
00:42:03,840 --> 00:42:12,560
do anything on the Power platform, Power FX is great to help you optimize flows, it's much faster,

319
00:42:12,560 --> 00:42:21,120
and it's also a lot cheaper. So it's just being able to manipulate data well, and I really love Jason

320
00:42:21,120 --> 00:42:31,840
now. Yeah, I hope the programming language, not the kid. No, not the person.

321
00:42:33,520 --> 00:42:44,720
Yeah, sorry. Yeah, when they look a little bit into the future, what did you think is

322
00:42:44,720 --> 00:42:56,960
the co-party studio, or I use more AI Foundry, will it cancel a lot of jobs and what skills I should

323
00:42:56,960 --> 00:43:05,920
deliver, then I know it was my job. How to build skills? That is the number one skill you should be able to do,

324
00:43:05,920 --> 00:43:14,320
and it's something that has never changed in humanity, I think, communication. Being able to

325
00:43:14,320 --> 00:43:19,760
communicate clearly, being able to say what you want to do, being able to, doesn't matter if you

326
00:43:19,760 --> 00:43:26,160
can draw like little diagram and stuff, as long as you're able to put it on paper, or in prompt,

327
00:43:26,160 --> 00:43:34,400
now in prompt box, to actually explain what you want to do. Communication is key. If you're not good at

328
00:43:34,400 --> 00:43:40,080
communication, it's going to be very hard for you to communicate effectively with the agents,

329
00:43:40,080 --> 00:43:45,280
what you want to do, and so on. Explain the use case you want to be able to build, that will improve

330
00:43:45,280 --> 00:43:54,160
efficiency, your job, and everything. And yeah, except that, AI Literacy is very important as well,

331
00:43:54,160 --> 00:43:59,920
not necessarily understanding how the model works and able to explain what goes on behind,

332
00:43:59,920 --> 00:44:05,920
but being able to use it. So being able to understand what an MCP is, what the skills are,

333
00:44:05,920 --> 00:44:14,160
able to trigger your workflows automatically, stuff like that, that really helps you a lot.

334
00:44:14,160 --> 00:44:20,800
Like my project manager, we got project managers, but I like to have a little AI

335
00:44:20,800 --> 00:44:25,200
project manager always by my side that does, for example, all the annoying things,

336
00:44:25,200 --> 00:44:31,040
like trying to find a meeting time with people. So are you available at that time? No, I'm available

337
00:44:31,040 --> 00:44:35,440
at that time. Oh, how about this time? And you get like, sometimes you get like 15 emails

338
00:44:35,440 --> 00:44:44,720
switching between you two. And I was like, yes, automate this, it's automated. So all this annoying

339
00:44:44,720 --> 00:44:52,640
things being able to kind of be entrepreneurial as well. And seeing, okay, that's a pain. And I

340
00:44:52,640 --> 00:45:05,040
really want to be able to improve this. And I really like the four piece. So like the four

341
00:45:05,040 --> 00:45:12,080
pieces with your pain and everything. So like the four piece in marketing, our product price,

342
00:45:12,080 --> 00:45:20,640
place and promotion, but it's more product price, pain and promotion for AI. So what you find the

343
00:45:20,640 --> 00:45:26,720
pain, you find the price, you find the right, sorry, you find the right product to build it in,

344
00:45:26,720 --> 00:45:32,800
you find the pricing and then you promote it. But everything goes from the pain and that's very

345
00:45:32,800 --> 00:45:42,000
much like entrepreneurship. Yeah, I know. But yeah, yeah, there's something you say is good.

346
00:45:42,000 --> 00:45:53,680
And shame on me, I never have used Copying Studio yet. I'm an AI Foundry guy. So there's something,

347
00:45:53,680 --> 00:46:06,560
a lot of new stuff like work, IQ, Copying agents and waters, skills and so on.

348
00:46:08,240 --> 00:46:18,000
What happened here? What's going on? So before work, IQ, you had like graph, and I'm sure a lot of

349
00:46:18,000 --> 00:46:26,720
people know of graph, graph is the API, you can query on the tenant. And it was, again, a single API,

350
00:46:26,720 --> 00:46:32,640
it wasn't packaged, it wasn't application intelligence. And work IQ kind of came and was like, we got

351
00:46:32,640 --> 00:46:40,480
MCPs, we got everything packaged nicely for AI is just plug and play. And that's what's going on.

352
00:46:40,480 --> 00:46:50,480
It's going and making the train AI ready for everything. Awesome. So I, I, I'll sometimes I,

353
00:46:50,480 --> 00:46:58,080
lose, I am a little bit overwhelmed with all these new Copilets, updates,

354
00:46:59,440 --> 00:47:10,880
what I do is only I say, okay, hey, Copilot, pet me in all the new, I photo, look, look at my

355
00:47:10,880 --> 00:47:23,120
link, look at my YouTube and make me 50 minutes. So yeah, I am really interesting, or I think it's

356
00:47:23,120 --> 00:47:31,040
really interesting in the future. Did you think will every business operate multi-agents?

357
00:47:31,040 --> 00:47:41,840
No, but I think every agent will at least operate one agent. And if not every business,

358
00:47:41,840 --> 00:47:47,840
at least part of the employees will use AI to help them in the day to day.

359
00:47:51,760 --> 00:47:57,840
And a lot of people still don't use AI in their daily life for starting to get used to it even now.

360
00:47:57,840 --> 00:48:11,360
And it is a good way to one see the capabilities, see what's the fun little bit, go, okay,

361
00:48:11,360 --> 00:48:15,520
create me a montage of my of my date for this weekend, I took pictures,

362
00:48:17,040 --> 00:48:22,800
user, user something to create sounds, etc. And you can have a lot of fun with it while learning.

363
00:48:22,800 --> 00:48:30,240
But people really need to get to get used to it, I think. And

364
00:48:30,240 --> 00:48:40,480
you don't need to go with Opus 5 or like the most expensive M-dash to actually get something out of it.

365
00:48:40,480 --> 00:48:46,720
Small agents like GPT 4.1 can already do quite a lot of summarizing and stuff like that.

366
00:48:47,280 --> 00:48:51,680
You don't need large language models always, small language models also very useful.

367
00:48:51,680 --> 00:48:58,720
If it's a little word flow where you need to summarize your emails coming in,

368
00:48:58,720 --> 00:49:05,760
you don't need to use something which costs like 15 or 20 dollars per token per million token.

369
00:49:05,760 --> 00:49:14,240
You can use GPT 5 mini, 5.15 mini, etc. I'm just get something small going,

370
00:49:15,040 --> 00:49:18,800
where it costs like I think confoundery was like 55 cents per million token.

371
00:49:18,800 --> 00:49:27,760
So yeah, really being able to know what you want to do, adapt it. And I think in the future,

372
00:49:27,760 --> 00:49:37,280
it's going, people are really going to use it a lot more. And especially agents are starting to

373
00:49:37,280 --> 00:49:41,840
email each other. I don't know if you've ever seen this, I'm starting to see it at work,

374
00:49:41,840 --> 00:49:49,200
especially for the meeting thing. And yeah, it's pretty funny because AI agents are organizing

375
00:49:49,200 --> 00:49:53,840
our calendar and just informing you, yeah, you got a meeting tomorrow. All right, cool. Thank you.

376
00:49:53,840 --> 00:50:03,040
Yeah, yeah, more destopness getting so expensive in the last time. I think more

377
00:50:03,600 --> 00:50:11,280
cordons, it's oh damn fuck, I have spent so many money. So I have two accounts on Azure,

378
00:50:11,280 --> 00:50:20,080
a one account, a sponsor or a star bank account with 150,000 Azure credits. And the other

379
00:50:20,080 --> 00:50:27,520
was 5,000, I say, okay, I think for cybersecurity, I have built a tool

380
00:50:28,240 --> 00:50:34,560
what should explain why the cybersecurity issues came from. So it's been up, I think,

381
00:50:34,560 --> 00:50:40,400
thousands, three hundred companies with need each penny, employees per company. And I have,

382
00:50:40,400 --> 00:50:50,480
I think I have there was not, I have take Claude, but not the newest version. And it's worked

383
00:50:50,480 --> 00:50:58,160
so hard. And after it's, after sometime is like, oh, your tokens are gone in two days,

384
00:50:58,640 --> 00:51:04,720
250,000, yeah, okay, Microsoft, hey, it's okay. But yeah, this was really hard. And then I have

385
00:51:04,720 --> 00:51:12,400
tried my own small, small language model. And let's more only, I think, six to no, to euro,

386
00:51:12,400 --> 00:51:22,160
I have spent so this is so amazing as a small or I say, I think the word, I don't like the word

387
00:51:22,160 --> 00:51:30,480
small language model, I like the more the word special language model because yeah, it's doing so,

388
00:51:30,480 --> 00:51:38,480
so great jobs. And especially in AI fondly, have we also, I don't know, I never was ago,

389
00:51:38,480 --> 00:51:46,560
I'm sorry for this, I'd better prepare, but is there also an option to use small language models

390
00:51:46,560 --> 00:51:53,680
in Copa? It's the way you got GPT 4.1. So you've got three billing levels in Copa studio,

391
00:51:53,680 --> 00:52:03,600
you've got default standard and premium. So like default is like 4.1. And then premium is like

392
00:52:03,600 --> 00:52:13,920
Opus or GPT 5 reasoning. And then standard is like GPT 5.5 chats or Sonnet 4.6. And yeah, you need

393
00:52:13,920 --> 00:52:18,400
to kind of juggle everything, make me sure you get the right model for a heavy workflow and stuff

394
00:52:18,400 --> 00:52:27,120
like that. And I think the, and on token consumption, I think the biggest mistake is kind of telling

395
00:52:27,120 --> 00:52:34,960
your agent to never call me, always try to do it until the end because it can like loop infinitely

396
00:52:34,960 --> 00:52:40,480
trying to do something. And I don't know if you've ever had this, but I woke up one morning with

397
00:52:41,200 --> 00:52:51,040
my 100 franc bill, which is like $800 on a Monday morning, right when I woke up. So yes, because I made

398
00:52:51,040 --> 00:52:55,120
that mistake to like, oh, it's fine, this is a simple workflow, we can do it easily.

399
00:52:55,120 --> 00:53:04,160
Yeah. So auto learning, auto healing and being able to say, I'm stuck in, in, in,

400
00:53:04,160 --> 00:53:09,600
in sends you an email or something, I'm going, you need to take a look at this, I'm stopping here

401
00:53:09,600 --> 00:53:17,920
because I'm like finding errors or something. This is a great way to, to not have those early morning

402
00:53:17,920 --> 00:53:27,200
bad days. This is a great, great way to start the morning with this kind of bill. So yeah.

403
00:53:27,200 --> 00:53:38,160
So okay, Microsoft supports reimburse me, but yeah, I love it. But I'm a co-pilot studio.

404
00:53:38,160 --> 00:53:47,360
How, how, when we talk about consumption and, and, and, yeah, the, the price is behind it.

405
00:53:47,360 --> 00:54:02,400
We have a lot of companies have a sea of all. How can we make him more happy? Is there any fine ops?

406
00:54:04,560 --> 00:54:10,640
fine, and co-pilot? You do have a lot of the Finops option in co-pilot studio,

407
00:54:10,640 --> 00:54:19,360
in co-pilot as well, in Foundry as well. It's kind of the, I think now you need agent 365 to do it.

408
00:54:19,360 --> 00:54:26,880
I haven't kept up to date too much lately with it. I'm waiting until it's a little bit more stable

409
00:54:26,880 --> 00:54:32,000
and more features come out to actually read everything, because it's still an evolution.

410
00:54:32,960 --> 00:54:39,840
But, yeah, you got, you got quite a few, quite a few tools. I think he got Finops in, in M365,

411
00:54:39,840 --> 00:54:44,720
in 365 admin.

412
00:54:44,720 --> 00:54:58,960
But then again, I really like to have a delegated app or an app registry, which is connected to,

413
00:54:58,960 --> 00:55:06,240
user subscription and everything where you're able to look at the information, get it directly

414
00:55:06,240 --> 00:55:12,880
live, query it also with an agent. That was actually was my, my last demo was AgentLands,

415
00:55:12,880 --> 00:55:19,200
where it's like connected to the whole cost management, so the billing subscription and everything,

416
00:55:19,200 --> 00:55:28,080
and was a reader for the user resource graph. So, and also the, the, the co-pilot packages.

417
00:55:28,720 --> 00:55:32,880
So, you're able to like list all the agents, see what they consume, see how many people are using it,

418
00:55:32,880 --> 00:55:39,840
and stuff like that. So, it's a little project I have now, unfortunately, kind of

419
00:55:39,840 --> 00:55:44,160
stalled because I don't have enough time for it, but it's something I really want to do for

420
00:55:44,160 --> 00:55:50,000
start of September and have it open source. You're going to need agent 365 subscription,

421
00:55:50,000 --> 00:55:55,760
because say, PIs are what you need it for, but so it's come to help you kind of interact with

422
00:55:55,760 --> 00:56:02,960
your, your environment, your GLP and stuff like that. So, kind of have everything in one place instead

423
00:56:02,960 --> 00:56:09,520
of going to clarity, preview, admin, this, admin, that, team's admin, co-pilot, admin, center, and so on.

424
00:56:09,520 --> 00:56:19,120
Really cool. What was the project? And then, yeah, the first project become a showcase. So,

425
00:56:19,920 --> 00:56:27,680
Microsoft had you a little bit fixed on the showcase stuff, or? Yeah, I like, well, with David Warner

426
00:56:27,680 --> 00:56:35,440
and his little community and with also the others, of course. But, no, I really liked, so it was the

427
00:56:35,440 --> 00:56:42,800
first Microsoft community I kind of found, and they accept me for demo. So, I'm very loyal in that

428
00:56:42,800 --> 00:56:51,280
way. And plus, I like to do demos behind the screen also, also in person, but it's much easier to kind of

429
00:56:51,280 --> 00:56:57,360
get to know how to do demos, how to talk behind the screen. So, for anybody listening, actually,

430
00:56:57,360 --> 00:57:05,760
wants to do demos. How do you recommend the platform community always enjoys having people do

431
00:57:05,760 --> 00:57:15,440
cool demos? Yeah, I think this is really, really, really cool. So, I have an every session,

432
00:57:15,440 --> 00:57:20,800
a rapid fire round. So, I ask short questions, and then you give more short answer.

433
00:57:20,800 --> 00:57:29,680
Ready? Let's go for it. TOR energy things during the development of a new agent.

434
00:57:31,040 --> 00:57:37,440
Gossi. When you have to talk to real people, teams or outlook?

435
00:57:37,440 --> 00:57:43,600
Teams. Your favorite co-pilot feature.

436
00:57:43,600 --> 00:57:56,960
Direct line integration. Oh, good. Your favorite power platform capability.

437
00:58:00,320 --> 00:58:06,560
Oh, that's the Uda Sardin. HTTP calls triggers on parot made.

438
00:58:06,560 --> 00:58:14,160
When the missbusters come to you and say, "Eliot, you are a part of the show, which

439
00:58:14,160 --> 00:58:16,000
miss will you bust?"

440
00:58:16,000 --> 00:58:28,960
That you cannot reduce AI agent tokens more than you can. You can reduce them a lot.

441
00:58:29,840 --> 00:58:35,920
Yeah. You're living in Switzerland. So, what's better? Shokunet or KS fondue?

442
00:58:35,920 --> 00:58:39,440
Oh, cheese fondue all the way.

443
00:58:39,440 --> 00:58:46,320
When Microsoft, I say, "Saturday night, I come to you and say, "Okay,

444
00:58:46,320 --> 00:58:52,160
"Eliot, you get all the money and resources to build the next feature on a co-pilot,

445
00:58:52,160 --> 00:58:54,640
studio. What will it be?"

446
00:58:55,200 --> 00:59:04,640
That the agent build itself.

447
00:59:04,640 --> 00:59:06,720
Okay.

448
00:59:06,720 --> 00:59:13,360
Are the skills and everything that really, you want to say the workflow and it builds it by itself

449
00:59:13,360 --> 00:59:18,160
on a whole other new level. It creates the skills. It goes and finds in other repos.

450
00:59:18,160 --> 00:59:22,880
It takes what's open source, subsorbs it and gives you a proposition.

451
00:59:23,280 --> 00:59:29,840
And the biggest lesson you have learned from Enterprise AI.

452
00:59:29,840 --> 00:59:40,560
It's hard to get people on board until you understand what motivates them.

453
00:59:40,560 --> 00:59:50,720
When you meet your younger yourself, I say, 20 years or so, what will he surprise most about you

454
00:59:50,720 --> 01:00:00,080
today? But I have access to internet. Unregulated on to net.

455
01:00:00,080 --> 01:00:16,720
Yeah, so cool for this. So, let's go to the closing and tell us a little bit about your

456
01:00:16,720 --> 01:00:21,840
your next project. You plan, you do a little bit deeper as before.

457
01:00:21,840 --> 01:00:29,040
After this podcast, I'm taking a long break. Well, it's a long break. But my next project is

458
01:00:29,040 --> 01:00:36,240
the agent, the agent in M365, a co-pilot chat, where it's able to help you manage the agents

459
01:00:36,240 --> 01:00:46,160
in your tenant for an admin. And so far, that sits taking a well-deserved break during the summer.

460
01:00:46,160 --> 01:00:48,240
That's one of your projects.

461
01:00:48,240 --> 01:00:58,720
Okay, finally, who showed the next guest here on the M365 event podcast?

462
01:00:58,720 --> 01:01:07,280
And what question showed I ask him, her, you don't afraid of?

463
01:01:07,280 --> 01:01:12,080
Do I get areas of expertise of the people?

464
01:01:13,360 --> 01:01:16,960
Sorry. Are areas of expertise of the people? Next people?

465
01:01:16,960 --> 01:01:23,360
Yeah, what secret question show? I ask you, you don't like to ask him.

466
01:01:23,360 --> 01:01:37,920
I don't like being mean. But I think a good question to ask is, how do you differentiate the

467
01:01:37,920 --> 01:01:43,840
different AI services? So like between agent builder, co-pilot studio, Foundry,

468
01:01:43,840 --> 01:01:51,600
like DSDKs and libraries, which are available. How do you pick which one to build on?

469
01:01:51,600 --> 01:01:54,960
And that's always a difficult one.

470
01:01:54,960 --> 01:01:58,960
What, what, what guest did you expect for this?

471
01:02:03,280 --> 01:02:09,280
That is a good question. Maybe, I don't know if you've already had Raffsahn on the call?

472
01:02:09,280 --> 01:02:16,960
Raffsahn, hi, oh, yeah, who's saying of?

473
01:02:16,960 --> 01:02:25,360
No, no, I had asked him some times, but I don't get in, actually, but I try it again.

474
01:02:25,360 --> 01:02:27,920
So I can say, Elliot, as a good guest you for you.

475
01:02:27,920 --> 01:02:30,640
Yeah?

476
01:02:30,640 --> 01:02:33,040
Yeah, so, yeah, Elliot.

477
01:02:33,040 --> 01:02:39,520
Thank you so much for joining me on Yemsa 65 Podcast and sharing your experience, especially in co-pilot,

478
01:02:39,520 --> 01:02:47,920
how you build multi-agent architecture, enterprise, and governance, and MCP integrations.

479
01:02:47,920 --> 01:02:54,480
Yeah, I think that's was really cool because we have also heard some real world experience,

480
01:02:54,480 --> 01:03:00,160
and not all the, yeah, the fantasy stuff. So thank you so much to being here.

481
01:03:01,040 --> 01:03:07,920
Thank you, Mirko, and if I may just have one last word, build little flows and parotemates to start

482
01:03:07,920 --> 01:03:14,560
and bring it up because you've got some great AI capabilities in parotemates.

483
01:03:14,560 --> 01:03:21,040
If you know how to do calls, HTTP calls, if you know how to integrate prompts, integrate them directly

484
01:03:21,040 --> 01:03:27,360
in your flows, analyze the things, try it, see how it goes out, send yourself messages in teams,

485
01:03:28,320 --> 01:03:36,320
don't get your building admins on your back, be slow, be cautious, but do enjoy, do test it,

486
01:03:36,320 --> 01:03:37,840
and that is all.

487
01:03:37,840 --> 01:03:44,880
Yeah, yeah, cool. So I think all the people find the links from you in the show notes, and yeah,

488
01:03:44,880 --> 01:03:49,200
thank you so much for being here. It was a real pleasure. I really enjoyed it.

489
01:03:49,200 --> 01:03:50,160
Thank you for having me.

490
01:03:50,160 --> 01:03:54,400
So much, my, all my face. It was really great. Thank you so much.

491
01:03:54,400 --> 01:03:56,560
Thank you very much for having me, Mirko.

492
01:03:57,120 --> 01:03:59,120
Have a great evening.