Inside Microsoft Foundry: Building the Next Generation of AI Apps with Jannik Reinhard [MVP]

1
00:00:00,000 --> 00:00:04,240
I come to another edition of the MC65FM podcast.

2
00:00:04,240 --> 00:00:07,040
Today we have the topic inside Microsoft Foundry,

3
00:00:07,040 --> 00:00:10,120
building the next generation of AI apps.

4
00:00:10,120 --> 00:00:12,840
My guest today is Janika Reinhardt,

5
00:00:12,840 --> 00:00:17,840
and we talk, yeah, we look at his amazing,

6
00:00:17,840 --> 00:00:19,520
amazing work.

7
00:00:19,520 --> 00:00:23,320
He has published 200 plus technical block posts

8
00:00:23,320 --> 00:00:27,060
and contribute to the open source community,

9
00:00:27,060 --> 00:00:31,260
delivering more than 30 public speaks every year

10
00:00:31,260 --> 00:00:32,460
across the world.

11
00:00:32,460 --> 00:00:36,260
And he also is the author of the book, "Modern Endpoint"

12
00:00:36,260 --> 00:00:38,940
and I, "I, "Driven IT."

13
00:00:38,940 --> 00:00:43,660
And, yeah, he has also built an architect

14
00:00:43,660 --> 00:00:47,740
in House Framework at BSF in, yeah,

15
00:00:47,740 --> 00:00:51,980
with 120,000 employees across multiple units, both.

16
00:00:51,980 --> 00:00:53,060
This is awesome.

17
00:00:53,060 --> 00:00:58,260
So, Janik, you have built this AI system by,

18
00:00:58,260 --> 00:01:01,740
that's used by more than 120,000 employees.

19
00:01:01,740 --> 00:01:04,060
What was the moment you realized AI

20
00:01:04,060 --> 00:01:06,900
would fundamentally change enterprise a T?

21
00:01:06,900 --> 00:01:10,540
Yeah, that's a good question in the end.

22
00:01:10,540 --> 00:01:13,340
I already worked with AI before.

23
00:01:13,340 --> 00:01:16,140
We have a big hype around it.

24
00:01:16,140 --> 00:01:19,860
I think the hype starts with the lounge of TETG-BT

25
00:01:19,860 --> 00:01:23,140
or maybe already with a GPT-3 model.

26
00:01:23,140 --> 00:01:25,580
I already worked before.

27
00:01:25,580 --> 00:01:29,420
Yeah, there was a hype of a large language model

28
00:01:29,420 --> 00:01:33,940
with a classical AI with an amaneet detection,

29
00:01:33,940 --> 00:01:36,260
forecasting stuff like this.

30
00:01:36,260 --> 00:01:40,220
This means I was perfectly prepared for the big hype

31
00:01:40,220 --> 00:01:43,740
and directly jump on the train with large language models

32
00:01:43,740 --> 00:01:47,620
and already developed here some small tools,

33
00:01:47,620 --> 00:01:50,620
I had some ideas which I can realize

34
00:01:50,620 --> 00:01:53,540
and therefore now on, yeah,

35
00:01:53,540 --> 00:01:55,940
I used that for multiple solutions I built,

36
00:01:55,940 --> 00:01:59,140
for sure I also used it every day for coding.

37
00:01:59,140 --> 00:02:04,140
That's crazy how much this enhanced your productivity

38
00:02:04,140 --> 00:02:08,860
and yeah, in the end, it's a lot of fun,

39
00:02:08,860 --> 00:02:13,860
but on the other hand, it's also very hard to be on track

40
00:02:13,860 --> 00:02:17,140
because so much changes every day.

41
00:02:17,140 --> 00:02:20,820
And yeah, to follow all these new features

42
00:02:20,820 --> 00:02:23,180
and all this stuff, that's hard.

43
00:02:23,180 --> 00:02:28,220
- There are two topics you're inside

44
00:02:28,220 --> 00:02:31,860
and we often would like a fight.

45
00:02:31,860 --> 00:02:35,060
It's the security part and the AI part.

46
00:02:35,060 --> 00:02:39,980
You're both MVP for security and for AI.

47
00:02:39,980 --> 00:02:44,980
How do these both worlds concepts work together today?

48
00:02:45,980 --> 00:02:49,980
- In the end, I think they work very well together

49
00:02:49,980 --> 00:02:51,980
because in my point of view,

50
00:02:51,980 --> 00:02:56,980
there are three technologies which are really essential

51
00:02:56,980 --> 00:02:57,980
for the future.

52
00:02:57,980 --> 00:02:59,980
It's data, it's AI and it's security.

53
00:02:59,980 --> 00:03:02,980
I think without strong security,

54
00:03:02,980 --> 00:03:05,980
you don't have to think about AI strategy

55
00:03:05,980 --> 00:03:07,980
because security is your insurance.

56
00:03:07,980 --> 00:03:10,980
Security is the most important piece

57
00:03:10,980 --> 00:03:12,980
what you should do is to make sure

58
00:03:12,980 --> 00:03:14,980
that you're the most important piece

59
00:03:14,980 --> 00:03:17,980
what you should consider in your IT landscape.

60
00:03:17,980 --> 00:03:21,980
AI for sure is the hype, but without good data,

61
00:03:21,980 --> 00:03:26,980
you also cannot build a good AI strategy.

62
00:03:26,980 --> 00:03:30,980
This means it's for me more multiple tier model,

63
00:03:30,980 --> 00:03:32,980
security data in AI.

64
00:03:32,980 --> 00:03:35,980
This means it works very well together,

65
00:03:35,980 --> 00:03:40,980
but yeah, to keep on track also here in both environments,

66
00:03:40,980 --> 00:03:42,980
it keeps me busy.

67
00:03:42,980 --> 00:03:43,980
- Awesome.

68
00:03:43,980 --> 00:03:47,980
And for people hearing Microsoft Foundry for the first time,

69
00:03:47,980 --> 00:03:50,980
how will you explain it in simple terms?

70
00:03:50,980 --> 00:03:52,980
- Yes.

71
00:03:52,980 --> 00:03:55,980
In the end, Foundry, on the first hand,

72
00:03:55,980 --> 00:03:59,980
it's a very easy way to deploy your models

73
00:03:59,980 --> 00:04:01,980
in a secure way.

74
00:04:01,980 --> 00:04:06,980
This means if you want to use TBT or of the AI models,

75
00:04:06,980 --> 00:04:09,980
if you want to use entropic models,

76
00:04:09,980 --> 00:04:12,980
you can deploy them on this platform.

77
00:04:12,980 --> 00:04:16,980
And you have hosting within the Microsoft Data Center

78
00:04:16,980 --> 00:04:19,980
with the complete security and trust deck

79
00:04:19,980 --> 00:04:21,980
what Microsoft provides you.

80
00:04:21,980 --> 00:04:24,980
This means encryption,

81
00:04:24,980 --> 00:04:26,980
put the models in a subnetwork,

82
00:04:26,980 --> 00:04:29,980
work with managed identities.

83
00:04:29,980 --> 00:04:32,980
In the end, in an enterprise environment,

84
00:04:32,980 --> 00:04:37,980
if you want to build LLM-powered apps,

85
00:04:37,980 --> 00:04:40,980
it's essential to use services like this.

86
00:04:40,980 --> 00:04:42,980
But on top of this,

87
00:04:42,980 --> 00:04:46,980
you have a lot of possibilities for observability

88
00:04:46,980 --> 00:04:51,980
to make sure that you are protected from the injection,

89
00:04:51,980 --> 00:04:56,980
from bad things,

90
00:04:56,980 --> 00:04:58,980
but you can do with these models,

91
00:04:58,980 --> 00:05:01,980
but also you have services to build agents,

92
00:05:01,980 --> 00:05:03,980
to build flows.

93
00:05:03,980 --> 00:05:08,980
This means it's made it very easy on a web interface

94
00:05:08,980 --> 00:05:10,980
to build your stuff,

95
00:05:10,980 --> 00:05:13,980
but also to do it via code,

96
00:05:13,980 --> 00:05:15,980
to have an SDK, for example,

97
00:05:15,980 --> 00:05:19,980
to build agents on a very easy way.

98
00:05:19,980 --> 00:05:23,980
And we have all these wonderful,

99
00:05:23,980 --> 00:05:26,980
large language models,

100
00:05:26,980 --> 00:05:29,980
all these machine learning stuff.

101
00:05:29,980 --> 00:05:31,980
What did you think?

102
00:05:31,980 --> 00:05:34,980
What did you think was different

103
00:05:34,980 --> 00:05:38,980
because machine learning and now AI?

104
00:05:38,980 --> 00:05:41,980
I think what we now have is

105
00:05:41,980 --> 00:05:43,980
in general purpose, AI.

106
00:05:43,980 --> 00:05:45,980
This means in the past,

107
00:05:45,980 --> 00:05:50,980
you have to very specialize your model for specific case.

108
00:05:50,980 --> 00:05:53,980
Maybe you have to pre-training it,

109
00:05:53,980 --> 00:05:57,980
and you have to invest a lot of work

110
00:05:57,980 --> 00:06:02,980
in this AI performance, yeah, very good.

111
00:06:02,980 --> 00:06:05,980
But today on your general purpose way,

112
00:06:05,980 --> 00:06:10,980
this means the large language models are very smart.

113
00:06:10,980 --> 00:06:12,980
They understand the context,

114
00:06:12,980 --> 00:06:17,980
they understand the full world around it.

115
00:06:17,980 --> 00:06:19,980
And in the most cases,

116
00:06:19,980 --> 00:06:21,980
it's all performs,

117
00:06:21,980 --> 00:06:24,980
everything what we know before.

118
00:06:24,980 --> 00:06:28,980
But sure, it's not a solution for everything.

119
00:06:28,980 --> 00:06:30,980
I think there are a lot of valid cases

120
00:06:30,980 --> 00:06:34,980
also for classical AI on machine learning.

121
00:06:34,980 --> 00:06:35,980
But in the end,

122
00:06:35,980 --> 00:06:38,980
it makes a lot of things very easy

123
00:06:38,980 --> 00:06:42,980
because it still works.

124
00:06:42,980 --> 00:06:45,980
What on your website?

125
00:06:45,980 --> 00:06:46,980
Epic Fusion.

126
00:06:46,980 --> 00:06:49,980
And I found this wonderful quote.

127
00:06:49,980 --> 00:06:52,980
IA isn't valuable because it's intelligent.

128
00:06:52,980 --> 00:06:55,980
It's valuable when it solves real business friction.

129
00:06:55,980 --> 00:06:59,980
Can you a little bit explain this quote?

130
00:06:59,980 --> 00:07:01,980
Yes.

131
00:07:01,980 --> 00:07:04,980
What I see is AI is a hype.

132
00:07:04,980 --> 00:07:07,980
And a lot of companies say,

133
00:07:07,980 --> 00:07:11,980
"Hey, we work with AI or we want to work with AI."

134
00:07:11,980 --> 00:07:17,980
But what I see is often they lack on use cases.

135
00:07:17,980 --> 00:07:21,980
They lack on a real concept or a real vision.

136
00:07:21,980 --> 00:07:24,980
What they want to do with AI.

137
00:07:24,980 --> 00:07:28,980
And the important thing is not jump on the hype train.

138
00:07:28,980 --> 00:07:32,980
It's more important to understand how the technology works,

139
00:07:32,980 --> 00:07:34,980
how your business works,

140
00:07:34,980 --> 00:07:37,980
and how you can use it to solve real issues,

141
00:07:37,980 --> 00:07:39,980
what you have,

142
00:07:39,980 --> 00:07:41,980
to make you more productive,

143
00:07:41,980 --> 00:07:43,980
to sell more products,

144
00:07:43,980 --> 00:07:45,980
to understand your processes,

145
00:07:45,980 --> 00:07:48,980
and check how you can make them stronger with AI.

146
00:07:48,980 --> 00:07:51,980
And see is companies have processes

147
00:07:51,980 --> 00:07:54,980
and try to put somewhere AI into it.

148
00:07:54,980 --> 00:07:57,980
But I think you have to rethink the whole world.

149
00:07:57,980 --> 00:08:01,980
It's more important to have an AI strategy

150
00:08:01,980 --> 00:08:04,980
and build your processes around it.

151
00:08:04,980 --> 00:08:09,980
I think that's the way how you can use all the scrumness,

152
00:08:09,980 --> 00:08:12,980
what this technology brings you,

153
00:08:12,980 --> 00:08:15,980
and bring the real value in your company also.

154
00:08:15,980 --> 00:08:18,980
And when you look in the future to make this,

155
00:08:18,980 --> 00:08:21,980
yeah, future ready,

156
00:08:21,980 --> 00:08:26,980
and not only to solve small issues what you have today.

157
00:08:26,980 --> 00:08:30,980
Yeah, when I look a little bit into LinkedIn,

158
00:08:30,980 --> 00:08:33,980
I think it's like an hype train.

159
00:08:33,980 --> 00:08:36,980
It's actually, it's AI copilot,

160
00:08:36,980 --> 00:08:39,980
but you focus heavily on agentec work roles.

161
00:08:39,980 --> 00:08:41,980
What does it differ?

162
00:08:41,980 --> 00:08:45,980
Yes, I think the term of it,

163
00:08:45,980 --> 00:08:48,980
agentec or agentec is,

164
00:08:48,980 --> 00:08:51,980
yeah, very different,

165
00:08:51,980 --> 00:08:56,980
integrated, if you reach for LinkedIn and all these blocks.

166
00:08:56,980 --> 00:09:00,980
A lot of people say also to a normal chat to be teached,

167
00:09:00,980 --> 00:09:03,980
I use an agent or stuff like this.

168
00:09:03,980 --> 00:09:05,980
In my understanding,

169
00:09:05,980 --> 00:09:08,980
an agent is someone where I can delegate work,

170
00:09:08,980 --> 00:09:11,980
and he works independent of me.

171
00:09:11,980 --> 00:09:13,980
And that's my goal.

172
00:09:13,980 --> 00:09:16,980
This means my goal is not to go in a chat and say,

173
00:09:16,980 --> 00:09:20,980
hey, please make a research on Google.

174
00:09:20,980 --> 00:09:22,980
What's the news from today?

175
00:09:22,980 --> 00:09:25,980
My goal is more to that an agent,

176
00:09:25,980 --> 00:09:28,980
understand my business, understand how I work.

177
00:09:28,980 --> 00:09:32,980
I enable them with different tools or possibilities

178
00:09:32,980 --> 00:09:35,980
that they, that he can really make actions

179
00:09:35,980 --> 00:09:39,980
and make me more productive and took over my work.

180
00:09:39,980 --> 00:09:43,980
That's, that's the important thing for me.

181
00:09:43,980 --> 00:09:45,980
And that's also my goal.

182
00:09:45,980 --> 00:09:49,980
My goal is really to build full code or broadcast solutions.

183
00:09:49,980 --> 00:09:54,980
We're very strong and agents with a good quality.

184
00:09:54,980 --> 00:09:57,980
I see a lot of also power platforms,

185
00:09:57,980 --> 00:10:00,980
so there are a lot of companies and citizens

186
00:10:00,980 --> 00:10:06,980
who have a lot of support for their business.

187
00:10:06,980 --> 00:10:10,980
So, that's the important thing for us.

188
00:10:10,980 --> 00:10:14,980
The fact is that it's not so much about the business.

189
00:10:14,980 --> 00:10:17,980
But it's not so much about the business.

190
00:10:17,980 --> 00:10:20,980
So, it's not so much about the business.

191
00:10:20,980 --> 00:10:22,980
But, it's not so much about the business.

192
00:10:22,980 --> 00:10:25,980
And then, I think that's the first thing that I do is,

193
00:10:25,980 --> 00:10:27,980
and I think,

194
00:10:27,980 --> 00:10:31,980
I think, I think, I don't want to talk about the technology.

195
00:10:31,980 --> 00:10:34,980
Yeah, make it to the start of this a bit.

196
00:10:34,980 --> 00:10:39,980
I think the technology is not longer the challenge.

197
00:10:39,980 --> 00:10:43,980
I think we have very strong models.

198
00:10:43,980 --> 00:10:45,980
We have good frameworks.

199
00:10:45,980 --> 00:10:48,980
The coding is very easy when you use AI.

200
00:10:48,980 --> 00:10:51,980
This means to bring an IT in reality.

201
00:10:51,980 --> 00:10:54,980
It's not a challenge.

202
00:10:54,980 --> 00:10:58,980
But, in the question, what is the real value?

203
00:10:58,980 --> 00:11:02,980
What is the real power behind?

204
00:11:02,980 --> 00:11:05,980
And for this, you need people in your company

205
00:11:05,980 --> 00:11:08,980
understand how your daily business work.

206
00:11:08,980 --> 00:11:12,980
Where are the points, which are really time-consuming?

207
00:11:12,980 --> 00:11:16,980
Where are the points, what you can automate?

208
00:11:16,980 --> 00:11:19,980
Or, can you take over from AI?

209
00:11:19,980 --> 00:11:21,980
And that's the point.

210
00:11:21,980 --> 00:11:25,980
The point is, if you have investment in change management,

211
00:11:25,980 --> 00:11:30,980
it's a lot of learning also that people with strong business knowledge

212
00:11:30,980 --> 00:11:34,980
also understand the technologies and the capabilities.

213
00:11:34,980 --> 00:11:39,980
I think that's something what you can't do from today to tomorrow.

214
00:11:39,980 --> 00:11:40,980
It's a journey.

215
00:11:40,980 --> 00:11:45,980
But, important is that you invest in change management,

216
00:11:45,980 --> 00:11:48,980
that you invest in your employee.

217
00:11:48,980 --> 00:11:51,980
And if you have, you have to invest in your employee.

218
00:11:51,980 --> 00:11:54,980
And you have to invest in your employee.

219
00:11:54,980 --> 00:11:57,980
And you have to invest in your employee.

220
00:11:57,980 --> 00:12:00,980
And you have to invest in your employee.

221
00:12:00,980 --> 00:12:03,980
You have to invest in your employee.

222
00:12:03,980 --> 00:12:06,980
You have to invest in your employee.

223
00:12:06,980 --> 00:12:07,980
You have to invest in your employee.

224
00:12:07,980 --> 00:12:10,980
You have to invest in your employee.

225
00:12:10,980 --> 00:12:12,980
You have to invest in your employee.

226
00:12:12,980 --> 00:12:14,980
You have to invest in your employee.

227
00:12:14,980 --> 00:12:16,980
You have to invest in your employee.

228
00:12:16,980 --> 00:12:18,980
You have to invest in your employee.

229
00:12:18,980 --> 00:12:20,980
You have to invest in your employee.

230
00:12:20,980 --> 00:12:22,980
You have to invest in your employee.

231
00:12:22,980 --> 00:12:23,980
You have to invest in your employee.

232
00:12:23,980 --> 00:12:25,980
You have to invest in your employee.

233
00:12:25,980 --> 00:12:26,980
You have to invest in your employee.

234
00:12:26,980 --> 00:12:27,980
You have to invest in your employee.

235
00:12:27,980 --> 00:12:28,980
You have to invest in your employee.

236
00:12:28,980 --> 00:12:29,980
You have to invest in your employee.

237
00:12:29,980 --> 00:12:30,980
You have to invest in your employee.

238
00:12:30,980 --> 00:12:31,980
You have to invest in your employee.

239
00:12:31,980 --> 00:12:32,980
You have to invest in your employee.

240
00:12:32,980 --> 00:12:33,980
You have to invest in your employee.

241
00:12:33,980 --> 00:12:34,980
You have to invest in your employee.

242
00:12:34,980 --> 00:12:35,980
You have to invest in your employee.

243
00:12:35,980 --> 00:12:36,980
You have to invest in your employee.

244
00:12:36,980 --> 00:12:37,980
You have to invest in your employee.

245
00:12:37,980 --> 00:12:38,980
You have to invest in your employee.

246
00:12:38,980 --> 00:12:39,980
You have to invest in your employee.

247
00:12:39,980 --> 00:12:40,980
You have to invest in your employee.

248
00:12:40,980 --> 00:12:41,980
You have to invest in your employee.

249
00:12:41,980 --> 00:12:42,980
You have to invest in your employee.

250
00:12:42,980 --> 00:12:44,980
You have to invest in your employee.

251
00:12:44,980 --> 00:12:46,980
You have to invest in your employee.

252
00:12:46,980 --> 00:12:48,980
You have to invest in your employee.

253
00:12:48,980 --> 00:12:49,980
You have to invest in your employee.

254
00:12:49,980 --> 00:12:50,980
You have to invest in your employee.

255
00:12:50,980 --> 00:12:51,980
You have to invest in your employee.

256
00:12:51,980 --> 00:12:52,980
You have to invest in your employee.

257
00:12:52,980 --> 00:12:53,980
You have to invest in your employee.

258
00:12:53,980 --> 00:12:54,980
You have to invest in your employee.

259
00:12:54,980 --> 00:12:55,980
You have to invest in your employee.

260
00:12:55,980 --> 00:12:56,980
You have to invest in your employee.

261
00:12:56,980 --> 00:12:57,980
You have to invest in your employee.

262
00:12:57,980 --> 00:12:58,980
You have to invest in your employee.

263
00:12:58,980 --> 00:12:59,980
You have to invest in your employee.

264
00:12:59,980 --> 00:13:00,980
You have to invest in your employee.

265
00:13:00,980 --> 00:13:01,980
You have to invest in your employee.

266
00:13:01,980 --> 00:13:02,980
You have to invest in your employee.

267
00:13:02,980 --> 00:13:03,980
You have to invest in your employee.

268
00:13:03,980 --> 00:13:04,980
You have to invest in your employee.

269
00:13:04,980 --> 00:13:05,980
You have to invest in your employee.

270
00:13:05,980 --> 00:13:06,980
You have to invest in your employee.

271
00:13:06,980 --> 00:13:07,980
You have to invest in your employee.

272
00:13:07,980 --> 00:13:09,980
You have to invest in your employee.

273
00:13:09,980 --> 00:13:11,980
You have to invest in your employee.

274
00:13:11,980 --> 00:13:13,980
You have to invest in your employee.

275
00:13:13,980 --> 00:13:14,980
You have to invest in your employee.

276
00:13:14,980 --> 00:13:16,980
You have to invest in your employee.

277
00:13:16,980 --> 00:13:17,980
You have to invest in your employee.

278
00:13:17,980 --> 00:13:18,980
You have to invest in your employee.

279
00:13:18,980 --> 00:13:19,980
You have to invest in your employee.

280
00:13:19,980 --> 00:13:20,980
You have to invest in your employee.

281
00:13:20,980 --> 00:13:21,980
You have to invest in your employee.

282
00:13:21,980 --> 00:13:22,980
You have to invest in your employee.

283
00:13:22,980 --> 00:13:23,980
You have to invest in your employee.

284
00:13:23,980 --> 00:13:24,980
You have to invest in your employee.

285
00:13:24,980 --> 00:13:25,980
You have to invest in your employee.

286
00:13:25,980 --> 00:13:26,980
You have to invest in your employee.

287
00:13:26,980 --> 00:13:27,980
You have to invest in your employee.

288
00:13:27,980 --> 00:13:28,980
You have to invest in your employee.

289
00:13:28,980 --> 00:13:29,980
You have to invest in your employee.

290
00:13:29,980 --> 00:13:30,980
You have to invest in your employee.

291
00:13:30,980 --> 00:13:31,980
You have to invest in your employee.

292
00:13:31,980 --> 00:13:32,980
You have to invest in your employee.

293
00:13:32,980 --> 00:13:33,980
You have to invest in your employee.

294
00:13:33,980 --> 00:13:35,980
You have to invest in your employee.

295
00:13:35,980 --> 00:13:36,980
You have to invest in your employee.

296
00:13:36,980 --> 00:13:37,980
You have to invest in your employee.

297
00:13:37,980 --> 00:13:39,980
You have to invest in your employee.

298
00:13:39,980 --> 00:13:40,980
You have to invest in your employee.

299
00:13:40,980 --> 00:13:41,980
You have to invest in your employee.

300
00:13:41,980 --> 00:13:42,980
You have to invest in your employee.

301
00:13:42,980 --> 00:13:43,980
You have to invest in your employee.

302
00:13:43,980 --> 00:13:44,980
You have to invest in your employee.

303
00:13:44,980 --> 00:13:45,980
You have to invest in your employee.

304
00:13:45,980 --> 00:13:46,980
You have to invest in your employee.

305
00:13:46,980 --> 00:13:47,980
You have to invest in your employee.

306
00:13:47,980 --> 00:13:48,980
You have to invest in your employee.

307
00:13:48,980 --> 00:13:49,980
You have to invest in your employee.

308
00:13:49,980 --> 00:13:50,980
You have to invest in your employee.

309
00:13:50,980 --> 00:13:51,980
You have to invest in your employee.

310
00:13:51,980 --> 00:13:52,980
You have to invest in your employee.

311
00:13:52,980 --> 00:13:53,980
You have to invest in your employee.

312
00:13:53,980 --> 00:13:54,980
You have to invest in your employee.

313
00:13:54,980 --> 00:13:55,980
You have to invest in your employee.

314
00:13:55,980 --> 00:13:56,980
You have to invest in your employee.

315
00:13:56,980 --> 00:13:57,980
You have to invest in your employee.

316
00:13:57,980 --> 00:13:59,980
You have to invest in your employee.

317
00:13:59,980 --> 00:14:01,980
You have to invest in your employee.

318
00:14:01,980 --> 00:14:03,980
You have to invest in your employee.

319
00:14:03,980 --> 00:14:04,980
You have to invest in your employee.

320
00:14:04,980 --> 00:14:05,980
You have to invest in your employee.

321
00:14:05,980 --> 00:14:06,980
You have to invest in your employee.

322
00:14:06,980 --> 00:14:07,980
You have to invest in your employee.

323
00:14:07,980 --> 00:14:08,980
You have to invest in your employee.

324
00:14:08,980 --> 00:14:09,980
You have to invest in your employee.

325
00:14:09,980 --> 00:14:10,980
You have to invest in your employee.

326
00:14:10,980 --> 00:14:11,980
You have to invest in your employee.

327
00:14:11,980 --> 00:14:12,980
You have to invest in your employee.

328
00:14:12,980 --> 00:14:13,980
You have to invest in your employee.

329
00:14:13,980 --> 00:14:14,980
You have to invest in your employee.

330
00:14:14,980 --> 00:14:15,980
You have to invest in your employee.

331
00:14:15,980 --> 00:14:16,980
You have to invest in your employee.

332
00:14:16,980 --> 00:14:17,980
You have to invest in your employee.

333
00:14:17,980 --> 00:14:18,980
You have to invest in your employee.

334
00:14:18,980 --> 00:14:19,980
You have to invest in your employee.

335
00:14:19,980 --> 00:14:20,980
You have to invest in your employee.

336
00:14:20,980 --> 00:14:21,980
You have to invest in your employee.

337
00:14:21,980 --> 00:14:22,980
You have to invest in your employee.

338
00:14:22,980 --> 00:14:24,980
You have to invest in your employee.

339
00:14:24,980 --> 00:14:26,980
You have to invest in your employee.

340
00:14:26,980 --> 00:14:28,980
You have to invest in your employee.

341
00:14:28,980 --> 00:14:29,980
You have to invest in your employee.

342
00:14:29,980 --> 00:14:30,980
You have to invest in your employee.

343
00:14:30,980 --> 00:14:31,980
You have to invest in your employee.

344
00:14:31,980 --> 00:14:32,980
You have to invest in your employee.

345
00:14:32,980 --> 00:14:33,980
You have to invest in your employee.

346
00:14:33,980 --> 00:14:34,980
You have to invest in your employee.

347
00:14:34,980 --> 00:14:35,980
You have to invest in your employee.

348
00:14:35,980 --> 00:14:36,980
You have to invest in your employee.

349
00:14:36,980 --> 00:14:37,980
You have to invest in your employee.

350
00:14:37,980 --> 00:14:38,980
You have to invest in your employee.

351
00:14:38,980 --> 00:14:39,980
You have to invest in your employee.

352
00:14:39,980 --> 00:14:40,980
You have to invest in your employee.

353
00:14:40,980 --> 00:14:41,980
You have to invest in your employee.

354
00:14:41,980 --> 00:14:42,980
You have to invest in your employee.

355
00:14:42,980 --> 00:14:43,980
You have to invest in your employee.

356
00:14:43,980 --> 00:14:44,980
You have to invest in your employee.

357
00:14:44,980 --> 00:14:45,980
You have to invest in your employee.

358
00:14:45,980 --> 00:14:46,980
You have to invest in your employee.

359
00:14:46,980 --> 00:14:47,980
You have to invest in your employee.

360
00:14:47,980 --> 00:14:49,980
You have to invest in your employee.

361
00:14:49,980 --> 00:14:51,980
You have to invest in your employee.

362
00:14:51,980 --> 00:14:53,980
You have to invest in your employee.

363
00:14:53,980 --> 00:14:54,980
You have to invest in your employee.

364
00:14:54,980 --> 00:14:55,980
You have to invest in your employee.

365
00:14:55,980 --> 00:14:56,980
You have to invest in your employee.

366
00:14:56,980 --> 00:14:57,980
You have to invest in your employee.

367
00:14:57,980 --> 00:14:58,980
You have to invest in your employee.

368
00:14:58,980 --> 00:14:59,980
You have to invest in your employee.

369
00:14:59,980 --> 00:15:00,980
You have to invest in your employee.

370
00:15:00,980 --> 00:15:01,980
You have to invest in your employee.

371
00:15:01,980 --> 00:15:02,980
You have to invest in your employee.

372
00:15:02,980 --> 00:15:03,980
You have to invest in your employee.

373
00:15:03,980 --> 00:15:04,980
You have to invest in your employee.

374
00:15:04,980 --> 00:15:05,980
You have to invest in your employee.

375
00:15:05,980 --> 00:15:06,980
You have to invest in your employee.

376
00:15:06,980 --> 00:15:07,980
You have to invest in your employee.

377
00:15:07,980 --> 00:15:08,980
You have to invest in your employee.

378
00:15:08,980 --> 00:15:09,980
You have to invest in your employee.

379
00:15:09,980 --> 00:15:10,980
You have to invest in your employee.

380
00:15:10,980 --> 00:15:11,980
You have to invest in your employee.

381
00:15:11,980 --> 00:15:12,980
You have to invest in your employee.

382
00:15:12,980 --> 00:15:14,980
You have to invest in your employee.

383
00:15:14,980 --> 00:15:16,980
You have to invest in your employee.

384
00:15:16,980 --> 00:15:18,980
You have to invest in your employee.

385
00:15:18,980 --> 00:15:19,980
You have to invest in your employee.

386
00:15:19,980 --> 00:15:20,980
You have to invest in your employee.

387
00:15:20,980 --> 00:15:21,980
You have to invest in your employee.

388
00:15:21,980 --> 00:15:22,980
You have to invest in your employee.

389
00:15:22,980 --> 00:15:23,980
You have to invest in your employee.

390
00:15:23,980 --> 00:15:24,980
You have to invest in your employee.

391
00:15:24,980 --> 00:15:25,980
You have to invest in your employee.

392
00:15:25,980 --> 00:15:26,980
You have to invest in your employee.

393
00:15:26,980 --> 00:15:27,980
You have to invest in your employee.

394
00:15:27,980 --> 00:15:28,980
You have to invest in your employee.

395
00:15:28,980 --> 00:15:29,980
You have to invest in your employee.

396
00:15:29,980 --> 00:15:30,980
You have to invest in your employee.

397
00:15:30,980 --> 00:15:31,980
You have to invest in your employee.

398
00:15:31,980 --> 00:15:32,980
You have to invest in your employee.

399
00:15:32,980 --> 00:15:33,980
You have to invest in your employee.

400
00:15:33,980 --> 00:15:34,980
You have to invest in your employee.

401
00:15:34,980 --> 00:15:35,980
You have to invest in your employee.

402
00:15:35,980 --> 00:15:36,980
You have to invest in your employee.

403
00:15:36,980 --> 00:15:37,980
You have to invest in your employee.

404
00:15:37,980 --> 00:15:39,980
You have to invest in your employee.

405
00:15:39,980 --> 00:15:41,980
You have to invest in your employee.

406
00:15:41,980 --> 00:15:43,980
You have to invest in your employee.

407
00:15:43,980 --> 00:15:44,980
You have to invest in your employee.

408
00:15:44,980 --> 00:15:45,980
You have to invest in your employee.

409
00:15:45,980 --> 00:15:46,980
You have to invest in your employee.

410
00:15:46,980 --> 00:15:47,980
You have to invest in your employee.

411
00:15:47,980 --> 00:15:48,980
You have to invest in your employee.

412
00:15:48,980 --> 00:15:49,980
You have to invest in your employee.

413
00:15:49,980 --> 00:15:50,980
You have to invest in your employee.

414
00:15:50,980 --> 00:15:51,980
You have to invest in your employee.

415
00:15:51,980 --> 00:15:52,980
You have to invest in your employee.

416
00:15:52,980 --> 00:15:53,980
You have to invest in your employee.

417
00:15:53,980 --> 00:15:54,980
You have to invest in your employee.

418
00:15:54,980 --> 00:15:55,980
You have to invest in your employee.

419
00:15:55,980 --> 00:15:56,980
You have to invest in your employee.

420
00:15:56,980 --> 00:15:57,980
You have to invest in your employee.

421
00:15:57,980 --> 00:15:58,980
You have to invest in your employee.

422
00:15:58,980 --> 00:15:59,980
You have to invest in your employee.

423
00:15:59,980 --> 00:16:00,980
You have to invest in your employee.

424
00:16:00,980 --> 00:16:01,980
You have to invest in your employee.

425
00:16:01,980 --> 00:16:02,980
You have to invest in your employee.

426
00:16:02,980 --> 00:16:04,980
You have to invest in your employee.

427
00:16:04,980 --> 00:16:05,980
You have to invest in your employee.

428
00:16:05,980 --> 00:16:06,980
You have to invest in your employee.

429
00:16:06,980 --> 00:16:08,980
You have to invest in your employee.

430
00:16:08,980 --> 00:16:09,980
You have to invest in your employee.

431
00:16:09,980 --> 00:16:10,980
You have to invest in your employee.

432
00:16:10,980 --> 00:16:11,980
You have to invest in your employee.

433
00:16:11,980 --> 00:16:12,980
You have to invest in your employee.

434
00:16:12,980 --> 00:16:13,980
You have to invest in your employee.

435
00:16:13,980 --> 00:16:14,980
You have to invest in your employee.

436
00:16:14,980 --> 00:16:15,980
You have to invest in your employee.

437
00:16:15,980 --> 00:16:16,980
You have to invest in your employee.

438
00:16:16,980 --> 00:16:17,980
You have to invest in your employee.

439
00:16:17,980 --> 00:16:18,980
You have to invest in your employee.

440
00:16:18,980 --> 00:16:19,980
You have to invest in your employee.

441
00:16:19,980 --> 00:16:20,980
You have to invest in your employee.

442
00:16:20,980 --> 00:16:21,980
You have to invest in your employee.

443
00:16:21,980 --> 00:16:22,980
You have to invest in your employee.

444
00:16:22,980 --> 00:16:23,980
You have to invest in your employee.

445
00:16:23,980 --> 00:16:24,980
You have to invest in your employee.

446
00:16:24,980 --> 00:16:25,980
You have to invest in your employee.

447
00:16:25,980 --> 00:16:26,980
You have to invest in your employee.

448
00:16:26,980 --> 00:16:28,980
You have to invest in your employee.

449
00:16:28,980 --> 00:16:29,980
You have to invest in your employee.

450
00:16:29,980 --> 00:16:30,980
You have to invest in your employee.

451
00:16:30,980 --> 00:16:31,980
You have to invest in your employee.

452
00:16:31,980 --> 00:16:32,980
You have to invest in your employee.

453
00:16:32,980 --> 00:16:33,980
You have to invest in your employee.

454
00:16:33,980 --> 00:16:34,980
You have to invest in your employee.

455
00:16:34,980 --> 00:16:35,980
You have to invest in your employee.

456
00:16:35,980 --> 00:16:36,980
You have to invest in your employee.

457
00:16:36,980 --> 00:16:37,980
You have to invest in your employee.

458
00:16:37,980 --> 00:16:38,980
You have to invest in your employee.

459
00:16:38,980 --> 00:16:39,980
You have to invest in your employee.

460
00:16:39,980 --> 00:16:40,980
You have to invest in your employee.

461
00:16:40,980 --> 00:16:41,980
You have to invest in your employee.

462
00:16:41,980 --> 00:16:42,980
You have to invest in your employee.

463
00:16:42,980 --> 00:16:43,980
You have to invest in your employee.

464
00:16:43,980 --> 00:16:44,980
You have to invest in your employee.

465
00:16:44,980 --> 00:16:45,980
You have to invest in your employee.

466
00:16:45,980 --> 00:16:46,980
You have to invest in your employee.

467
00:16:46,980 --> 00:16:47,980
You have to invest in your employee.

468
00:16:47,980 --> 00:16:48,980
You have to invest in your employee.

469
00:16:48,980 --> 00:16:49,980
You have to invest in your employee.

470
00:16:49,980 --> 00:16:51,980
You have to invest in your employee.

471
00:16:51,980 --> 00:16:53,980
You have to invest in your employee.

472
00:16:53,980 --> 00:16:55,980
You have to invest in your employee.

473
00:16:55,980 --> 00:16:57,980
You have to invest in your employee.

474
00:16:57,980 --> 00:16:59,980
You have to invest in your employee.

475
00:16:59,980 --> 00:17:01,980
You have to invest in your employee.

476
00:17:01,980 --> 00:17:02,980
You have to invest in your employee.

477
00:17:02,980 --> 00:17:03,980
You have to invest in your employee.

478
00:17:03,980 --> 00:17:04,980
You have to invest in your employee.

479
00:17:04,980 --> 00:17:05,980
You have to invest in your employee.

480
00:17:05,980 --> 00:17:06,980
You have to invest in your employee.

481
00:17:06,980 --> 00:17:07,980
You have to invest in your employee.

482
00:17:07,980 --> 00:17:08,980
You have to invest in your employee.

483
00:17:08,980 --> 00:17:09,980
You have to invest in your employee.

484
00:17:09,980 --> 00:17:10,980
You have to invest in your employee.

485
00:17:10,980 --> 00:17:11,980
You have to invest in your employee.

486
00:17:11,980 --> 00:17:12,980
You have to invest in your employee.

487
00:17:12,980 --> 00:17:13,980
You have to invest in your employee.

488
00:17:13,980 --> 00:17:14,980
You have to invest in your employee.

489
00:17:14,980 --> 00:17:15,980
You have to invest in your employee.

490
00:17:15,980 --> 00:17:16,980
You have to invest in your employee.

491
00:17:16,980 --> 00:17:17,980
You have to invest in your employee.

492
00:17:17,980 --> 00:17:19,980
You have to invest in your employee.

493
00:17:19,980 --> 00:17:21,980
You have to invest in your employee.

494
00:17:21,980 --> 00:17:23,980
You have to invest in your employee.

495
00:17:23,980 --> 00:17:25,980
You have to invest in your employee.

496
00:17:25,980 --> 00:17:27,980
You have to invest in your employee.

497
00:17:27,980 --> 00:17:29,980
You have to invest in your employee.

498
00:17:29,980 --> 00:17:30,980
You have to invest in your employee.

499
00:17:30,980 --> 00:17:31,980
You have to invest in your employee.

500
00:17:31,980 --> 00:17:32,980
You have to invest in your employee.

501
00:17:32,980 --> 00:17:33,980
You have to invest in your employee.

502
00:17:33,980 --> 00:17:34,980
You have to invest in your employee.

503
00:17:34,980 --> 00:17:35,980
You have to invest in your employee.

504
00:17:35,980 --> 00:17:36,980
You have to invest in your employee.

505
00:17:36,980 --> 00:17:37,980
You have to invest in your employee.

506
00:17:37,980 --> 00:17:38,980
You have to invest in your employee.

507
00:17:38,980 --> 00:17:39,980
You have to invest in your employee.

508
00:17:39,980 --> 00:17:40,980
You have to invest in your employee.

509
00:17:40,980 --> 00:17:41,980
You have to invest in your employee.

510
00:17:41,980 --> 00:17:42,980
You have to invest in your employee.

511
00:17:42,980 --> 00:17:43,980
You have to invest in your employee.

512
00:17:43,980 --> 00:17:44,980
You have to invest in your employee.

513
00:17:44,980 --> 00:17:45,980
You have to invest in your employee.

514
00:17:45,980 --> 00:17:47,980
You have to invest in your employee.

515
00:17:47,980 --> 00:17:49,980
You have to invest in your employee.

516
00:17:49,980 --> 00:17:51,980
You have to invest in your employee.

517
00:17:51,980 --> 00:17:53,980
You have to invest in your employee.

518
00:17:53,980 --> 00:17:55,980
You have to invest in your employee.

519
00:17:55,980 --> 00:17:57,980
You have to invest in your employee.

520
00:17:57,980 --> 00:17:58,980
You have to invest in your employee.

521
00:17:58,980 --> 00:17:59,980
You have to invest in your employee.

522
00:17:59,980 --> 00:18:00,980
You have to invest in your employee.

523
00:18:00,980 --> 00:18:01,980
You have to invest in your employee.

524
00:18:01,980 --> 00:18:02,980
You have to invest in your employee.

525
00:18:02,980 --> 00:18:03,980
You have to invest in your employee.

526
00:18:03,980 --> 00:18:04,980
You have to invest in your employee.

527
00:18:04,980 --> 00:18:05,980
You have to invest in your employee.

528
00:18:05,980 --> 00:18:06,980
You have to invest in your employee.

529
00:18:06,980 --> 00:18:07,980
You have to invest in your employee.

530
00:18:07,980 --> 00:18:08,980
You have to invest in your employee.

531
00:18:08,980 --> 00:18:09,980
You have to invest in your employee.

532
00:18:09,980 --> 00:18:10,980
You have to invest in your employee.

533
00:18:10,980 --> 00:18:11,980
You have to invest in your employee.

534
00:18:11,980 --> 00:18:12,980
You have to invest in your employee.

535
00:18:12,980 --> 00:18:13,980
You have to invest in your employee.

536
00:18:13,980 --> 00:18:15,980
You have to invest in your employee.

537
00:18:15,980 --> 00:18:17,980
You have to invest in your employee.

538
00:18:17,980 --> 00:18:19,980
You have to invest in your employee.

539
00:18:19,980 --> 00:18:20,980
You have to invest in your employee.

540
00:18:20,980 --> 00:18:22,980
You have to invest in your employee.

541
00:18:22,980 --> 00:18:23,980
You have to invest in your employee.

542
00:18:23,980 --> 00:18:24,980
You have to invest in your employee.

543
00:18:24,980 --> 00:18:25,980
You have to invest in your employee.

544
00:18:25,980 --> 00:18:26,980
You have to invest in your employee.

545
00:18:26,980 --> 00:18:27,980
You have to invest in your employee.

546
00:18:27,980 --> 00:18:28,980
You have to invest in your employee.

547
00:18:28,980 --> 00:18:29,980
You have to invest in your employee.

548
00:18:29,980 --> 00:18:30,980
You have to invest in your employee.

549
00:18:30,980 --> 00:18:31,980
You have to invest in your employee.

550
00:18:31,980 --> 00:18:32,980
You have to invest in your employee.

551
00:18:32,980 --> 00:18:33,980
You have to invest in your employee.

552
00:18:33,980 --> 00:18:34,980
You have to invest in your employee.

553
00:18:34,980 --> 00:18:35,980
You have to invest in your employee.

554
00:18:35,980 --> 00:18:36,980
You have to invest in your employee.

555
00:18:36,980 --> 00:18:37,980
You have to invest in your employee.

556
00:18:37,980 --> 00:18:38,980
You have to invest in your employee.

557
00:18:38,980 --> 00:18:39,980
You have to invest in your employee.

558
00:18:39,980 --> 00:18:41,980
You have to invest in your employee.

559
00:18:41,980 --> 00:18:43,980
You have to invest in your employee.

560
00:18:43,980 --> 00:18:45,980
You have to invest in your employee.

561
00:18:45,980 --> 00:18:46,980
You have to invest in your employee.

562
00:18:46,980 --> 00:18:47,980
You have to invest in your employee.

563
00:18:47,980 --> 00:18:48,980
You have to invest in your employee.

564
00:18:48,980 --> 00:18:49,980
You have to invest in your employee.

565
00:18:49,980 --> 00:18:50,980
You have to invest in your employee.

566
00:18:50,980 --> 00:18:51,980
You have to invest in your employee.

567
00:18:51,980 --> 00:18:52,980
You have to invest in your employee.

568
00:18:52,980 --> 00:18:53,980
You have to invest in your employee.

569
00:18:53,980 --> 00:18:54,980
You have to invest in your employee.

570
00:18:54,980 --> 00:18:55,980
You have to invest in your employee.

571
00:18:55,980 --> 00:18:56,980
You have to invest in your employee.

572
00:18:56,980 --> 00:18:57,980
You have to invest in your employee.

573
00:18:57,980 --> 00:18:58,980
You have to invest in your employee.

574
00:18:58,980 --> 00:18:59,980
You have to invest in your employee.

575
00:18:59,980 --> 00:19:00,980
You have to invest in your employee.

576
00:19:00,980 --> 00:19:01,980
You have to invest in your employee.

577
00:19:01,980 --> 00:19:02,980
You have to invest in your employee.

578
00:19:02,980 --> 00:19:03,980
You have to invest in your employee.

579
00:19:03,980 --> 00:19:04,980
You have to invest in your employee.

580
00:19:04,980 --> 00:19:06,980
You have to invest in your employee.

581
00:19:06,980 --> 00:19:08,980
You have to invest in your employee.

582
00:19:08,980 --> 00:19:10,980
You have to invest in your employee.

583
00:19:10,980 --> 00:19:11,980
You have to invest in your employee.

584
00:19:11,980 --> 00:19:12,980
You have to invest in your employee.

585
00:19:12,980 --> 00:19:13,980
You have to invest in your employee.

586
00:19:13,980 --> 00:19:14,980
You have to invest in your employee.

587
00:19:14,980 --> 00:19:15,980
You have to invest in your employee.

588
00:19:15,980 --> 00:19:16,980
You have to invest in your employee.

589
00:19:16,980 --> 00:19:17,980
You have to invest in your employee.

590
00:19:17,980 --> 00:19:18,980
You have to invest in your employee.

591
00:19:18,980 --> 00:19:19,980
You have to invest in your employee.

592
00:19:19,980 --> 00:19:20,980
You have to invest in your employee.

593
00:19:20,980 --> 00:19:21,980
You have to invest in your employee.

594
00:19:21,980 --> 00:19:22,980
You have to invest in your employee.

595
00:19:22,980 --> 00:19:23,980
You have to invest in your employee.

596
00:19:23,980 --> 00:19:24,980
You have to invest in your employee.

597
00:19:24,980 --> 00:19:25,980
You have to invest in your employee.

598
00:19:25,980 --> 00:19:26,980
You have to invest in your employee.

599
00:19:26,980 --> 00:19:27,980
You have to invest in your employee.

600
00:19:27,980 --> 00:19:28,980
You have to invest in your employee.

601
00:19:28,980 --> 00:19:29,980
You have to invest in your employee.

602
00:19:29,980 --> 00:19:33,980
I'm a big fan to not use frameworks because frameworks often lack

603
00:19:33,980 --> 00:19:41,980
And I'm personally I'm fast when I build it on scratch.

604
00:19:41,980 --> 00:19:44,980
But if you are starting to work in,

605
00:19:44,980 --> 00:19:47,980
yeah, working or building AIs.

606
00:19:47,980 --> 00:19:51,980
It makes it very easy to install the SDK.

607
00:19:51,980 --> 00:19:55,980
You can interact with your models also from every vendor.

608
00:19:55,980 --> 00:20:01,980
You don't have to, as I said, take care about data loss.

609
00:20:01,980 --> 00:20:05,980
And the beginning for someone who is new is very easy.

610
00:20:05,980 --> 00:20:09,980
Also, you can click through the UI and agent.

611
00:20:09,980 --> 00:20:14,980
You can connect it with your mcp or connect it to the AI search

612
00:20:14,980 --> 00:20:16,980
or foundry IQ.

613
00:20:16,980 --> 00:20:19,980
You can connect it to fabric IQ.

614
00:20:19,980 --> 00:20:24,980
This means you can crown it on a very large scale.

615
00:20:24,980 --> 00:20:27,980
You can crown it on a very easy way with your business data.

616
00:20:27,980 --> 00:20:30,980
You can index your documents on the search.

617
00:20:30,980 --> 00:20:34,980
You can make a rack without developing a rack.

618
00:20:34,980 --> 00:20:40,980
This means it makes it especially for beginners.

619
00:20:40,980 --> 00:20:48,980
Very easy to build full-code applications or agents without deep knowledge.

620
00:20:48,980 --> 00:20:54,980
And what, thank you, which capability inside foundry

621
00:20:54,980 --> 00:20:58,980
are most underrated, actually?

622
00:20:58,980 --> 00:21:06,980
Let me think about what is underrated.

623
00:21:06,980 --> 00:21:11,980
Yeah, as I said, it's more or less the full content filter,

624
00:21:11,980 --> 00:21:14,980
prompt interactions, all this stuff,

625
00:21:14,980 --> 00:21:18,980
because I can configure it as I want to have it.

626
00:21:18,980 --> 00:21:22,980
Also, things to connect it with per view.

627
00:21:22,980 --> 00:21:30,980
Also, this is very strong to make sure that not sensible data will be exposed to model.

628
00:21:30,980 --> 00:21:38,980
All these things where I can click in minutes through a visit

629
00:21:38,980 --> 00:21:41,980
for me, that's underrated and that's something.

630
00:21:41,980 --> 00:21:45,980
What other providers like OpenAI and Tropics cannot deliver

631
00:21:45,980 --> 00:21:50,980
because they have not this strong ecosystem, what Microsoft has?

632
00:21:50,980 --> 00:21:58,980
When you think foundry, I think it's enterprise level

633
00:21:58,980 --> 00:22:06,980
or can also you small or mid-sized companies rebuild an app today.

634
00:22:06,980 --> 00:22:14,980
I would recommend it for every enterprise. This means independent of small or large size.

635
00:22:14,980 --> 00:22:19,980
In the end, I think the most underpriced is hopefully run.

636
00:22:19,980 --> 00:22:24,980
In the end, they deploy their devices within June.

637
00:22:24,980 --> 00:22:29,980
This means they already have the environment, the edge attendant.

638
00:22:29,980 --> 00:22:34,980
And to start with, it's a question of two minutes.

639
00:22:34,980 --> 00:22:41,980
You go deploy resource, foundry, go to deployment, deploy your model, and you're good to go.

640
00:22:41,980 --> 00:22:44,980
And you can start with.

641
00:22:44,980 --> 00:22:48,980
And this is also what I can encourage everyone to not wait.

642
00:22:48,980 --> 00:22:57,980
And let's overall from the technologies and also you're competitive, don't sleep.

643
00:22:57,980 --> 00:23:00,980
Really start, try it out.

644
00:23:00,980 --> 00:23:09,980
Bring your ideas into reality. And as I said, when you have AI coding tools like Lord or CodeX,

645
00:23:09,980 --> 00:23:13,980
that's so easy. You have your idea, you type in what you want to do,

646
00:23:13,980 --> 00:23:22,980
and you came from idea to a product, not a product, but to POC in minutes.

647
00:23:22,980 --> 00:23:27,980
Yeah, can you walk us through a real world architecture of an AI app you have built?

648
00:23:27,980 --> 00:23:30,980
Usually Microsoft technologies?

649
00:23:30,980 --> 00:23:37,980
Yes, what I mostly do is I work with an app service.

650
00:23:37,980 --> 00:23:41,980
I build my application as a container.

651
00:23:41,980 --> 00:23:46,980
I deploy this container via a GitHub pipeline into this app service.

652
00:23:46,980 --> 00:23:52,980
I connect this app service with network to foundry and use the model.

653
00:23:52,980 --> 00:23:59,980
This means that's the most cases sufficient. You have your application logic and if you have your model.

654
00:23:59,980 --> 00:24:09,980
And if you need to start some data, you deploy also on SQL database or sometimes also on a table storage is sufficient.

655
00:24:09,980 --> 00:24:18,980
This means the architecture is sometimes not so advanced.

656
00:24:18,980 --> 00:24:23,980
This means an app service and foundry is the most cases sufficient.

657
00:24:23,980 --> 00:24:33,980
What I also really like is the AI search in my condufutes, one of the simplest and best search engines, which I try so far.

658
00:24:33,980 --> 00:24:43,980
This means to get a ton of data index and you can retrieve them in a very good quality.

659
00:24:43,980 --> 00:24:55,980
This means if you want to build a record application with AI search, it's no brainer and the quality is good only the price to get started is not as low.

660
00:24:55,980 --> 00:25:08,980
But I think in the meantime, there's also free tier to get up to three indexes and I think for building in POC, it's no more cases sufficient.

661
00:25:08,980 --> 00:25:17,980
What roles do vector databases, memory and grounding play and modern AI apps today?

662
00:25:17,980 --> 00:25:23,980
Depends for sure on which app you work, but I think memory is in peace, which is very important.

663
00:25:23,980 --> 00:25:40,980
Especially when you want to build agents, which also remember what they already did or when you have a chat application, you type something in and you're not will give this context again and again.

664
00:25:40,980 --> 00:25:52,980
I think memory is very important also rack is no most cases also very important to retrieve knowledge and also here in AI.

665
00:25:52,980 --> 00:25:56,980
And AI search for possibility to have also a Chandicrack.

666
00:25:56,980 --> 00:26:06,980
This means not only have a question, David in the search, retrieve the results and generate the answer with a Chandicrack you have multi turn approach.

667
00:26:06,980 --> 00:26:18,980
This means they agent understand your question and go with multiple turns through the data with different queries to retrieve really what you need to answer your question.

668
00:26:18,980 --> 00:26:28,980
And yeah, I think today on I also publish a blog this week about it, you're so much terms of skills, you have mcp or frag.

669
00:26:28,980 --> 00:26:36,980
And this means to came for a turn solution, you often have multiple tools, but you can use.

670
00:26:36,980 --> 00:26:51,980
And to get the same results, the question is always what's the best way also my opinion mcp has a lot of disadvantages.

671
00:26:51,980 --> 00:27:16,980
And for me, CLI is the most cases the better way and to use to use or to give the models the possibility to execute code and then yeah right CLI commands to interact with tools, but what I also did I am a current role at the big fusion I build big AI platform.

672
00:27:16,980 --> 00:27:33,980
And what I did here is I use the advantage or the good things from mcp but build my own standard which has not the disadvantageous what mcp has like to provide too much context to the models.

673
00:27:33,980 --> 00:27:46,980
For example, when you have 300 tools, you are often out of the context window and it's not good to give too much context to model because then they don't find the middle in the haystack.

674
00:27:46,980 --> 00:27:56,980
And that's the point to have here good balance and all these things are also orchestration to came back to your to your previous question.

675
00:27:56,980 --> 00:28:08,980
For sure you can build the application you can give their all tools a lot of mcp server and hope that the model find find the right way.

676
00:28:08,980 --> 00:28:21,980
But also if you build a good orchestration to figure out what's in way to give only the context which is really needed, but all the possibilities that the agent can be.

677
00:28:21,980 --> 00:28:33,980
Yeah, fulfill the needs. I think that's that's more more less a challenge and that makes a difference between a good orchestration and a bad orchestration.

678
00:28:33,980 --> 00:28:34,980
Awesome.

679
00:28:34,980 --> 00:28:47,980
When we think a little bit about the topics compliance and governance, can you solve some tips, how do you work with it or bring it to scale?

680
00:28:47,980 --> 00:29:04,980
Yeah, I think governance in the environment of AI is essential and what I see is that a lot of companies not yet deal with it, but it's important.

681
00:29:04,980 --> 00:29:19,980
I think it's important to not wait before you implement something because you should be quick, you should try it out, you should collect experience, but don't forget governance.

682
00:29:19,980 --> 00:29:48,980
And also here I think what Microsoft delivers so far, I'm quite happy. I think now with e7 and h365, you have a good way to inventory, toy your agents, you have a good possibility to also give them identity to bring them into a zero trust approach to give your agents permissions on on specific tools or apps.

683
00:29:48,980 --> 00:30:06,980
What you have, you have also, I think also this is really really underrated global secure access with global secure access and per view, you can also make sure that you don't expose.

684
00:30:06,980 --> 00:30:20,980
You have sensitive data from your company, you can also make sure when someone used a chat GBT that know data is exposed and I think the stack what we have is quite good.

685
00:30:20,980 --> 00:30:43,980
So, yeah, create a concept for full governance concept from co pilot to full code application. This means everything what we have here and that's important to consider this and make here a good concept for your company that the developer, yeah, are protected to make too much mistakes.

686
00:30:43,980 --> 00:30:53,980
Yeah, this is awesome. And then we look a little bit, I think, governance wise risk are there is no topic is.

687
00:30:53,980 --> 00:31:03,980
I see actually through AI, we have a chance to bring this more to companies that you see also.

688
00:31:03,980 --> 00:31:23,980
You mean to bring compliance and security more to company topic, I think that's the chance for for I think we have start with data and also OK, it's it's more there some developer use, I don't know.

689
00:31:23,980 --> 00:31:33,980
I think that's a great question. And now everyone use it. So I think that's changed the mind how company think about this topics.

690
00:31:33,980 --> 00:31:43,980
Yes, so yeah, you're completely right. And I think with AI, a lot of companies wake up and also see what they missed in the last years.

691
00:31:43,980 --> 00:31:56,980
And then we have a company who still run on premise and now you have the challenge to bring first everything to the cloud before you can think about AI and that are all challenges which now pops up.

692
00:31:56,980 --> 00:32:11,980
This means if you have, yeah, good concepts about how you run your application, the cloud about compliance about security already in the past and you can now directly jump on the train with AI.

693
00:32:11,980 --> 00:32:27,980
And that's a big advantage, but for sure, for now it's also door open up a lot of companies to convince their management to also go this pass and invest more into data into security into move to cloud.

694
00:32:27,980 --> 00:32:45,980
Because that's now essential to survive for sure, not every company can go in the cloud, regulated environments, they have it, yeah, have much harder life and also they are possibilities to run your models on premise and so on.

695
00:32:45,980 --> 00:32:59,980
All these frontier models and all the capabilities and also the strong, the strong models, I think, are running in a cloud and the AI world mostly lives in a cloud.

696
00:32:59,980 --> 00:33:11,980
This means, yeah, it's good when you already go this pass before the AI times.

697
00:33:11,980 --> 00:33:25,980
So what did you think is the AI hype cycle helping innovation or creating unrealistic exceptions, but I think a little bit about the start of co pilot for my perspective.

698
00:33:25,980 --> 00:33:41,980
This is cool, I can create bar points and can do my outlook emails and for me, nothing works like I think, what's your, what did you think?

699
00:33:41,980 --> 00:33:51,980
As I said, the technology is here and in my point of view, also as of today, you can solve, I would say, 99% of the cases.

700
00:33:51,980 --> 00:33:59,980
But the marketing and the expectation what a lot of vendors do is far away from the capabilities.

701
00:33:59,980 --> 00:34:17,980
This means what Microsoft did at the beginning, they had marketing and your expectation was so high and when you tried out, you was very, yeah, and this annoyed what they really deliver.

702
00:34:17,980 --> 00:34:27,980
At the beginning, co pilot was in the most cases not useful, also today of a lot of cases where co pilot cannot help you, also that you expected.

703
00:34:27,980 --> 00:34:33,980
But you see it will get better and better.

704
00:34:33,980 --> 00:34:55,980
On the other hand, I see also the strategy of some vendors like service now, SAP and so on and so forth, what, who think they can, yeah, deliver a strategy, which are only in their shell.

705
00:34:55,980 --> 00:35:05,980
This means you buy additional licenses and you only run everything into their platform and that's the complete wrong approach.

706
00:35:05,980 --> 00:35:19,980
I think to be successful here, you should be open, you should provide interfaces in the AI, I should run where the customer has their platform or their environment and not inside the tools.

707
00:35:19,980 --> 00:35:39,980
I think it's kind of co pilot to solve small challenges like answer email for sure should be in the tool, but you should also be open to expose your interfaces and your data, which in the end belong to the customer also to them that they can build their own solutions.

708
00:35:39,980 --> 00:36:03,980
And as I said, to answer your questions, everything is possible as of today. If you are smart, if you build your own solutions, but often what the vendors promise you are far away from what they really deliver you, they often sell their roadmap and not their as of today state.

709
00:36:03,980 --> 00:36:11,980
And when we look a little bit and foundry, we can use all these wonderful models.

710
00:36:11,980 --> 00:36:19,980
And I think I found the premium models sometimes really expensive.

711
00:36:19,980 --> 00:36:37,980
Is there an option or not, there is the option there, but how can companies or developers choose between models and also did they every time need a I-mortals or can they also use something cheaper?

712
00:36:37,980 --> 00:36:45,980
I don't know, a vector search from error or makes it sensible nearly all layer.

713
00:36:45,980 --> 00:37:01,980
Yeah, I think there are a lot of different possibilities, how you can save tokens and with is also costs or how you can lower the model or the quality of model.

714
00:37:01,980 --> 00:37:17,980
And that means on the first hand, if you have a good orchestration, you can also save costs because as less turns or less context, you have to give to model to generate a good answer is more talking to safe and with this you can also lower the cost.

715
00:37:17,980 --> 00:37:45,980
This is one point also if you can build a very good orchestration where also you have from high value context, you don't need to oppose to for every task you can also go with high quality or GPT five for many for example, this means and that's also what differs and good developer to invite co to understand how things works.

716
00:37:45,980 --> 00:38:03,980
And how you can use the capabilities, what the model delivers you on an efficient way and that's I think also important points to figure out and that's also kind of experience or feeling what you need.

717
00:38:03,980 --> 00:38:19,980
What's now for this task right model and how can I feed it on an efficient way and that's something what yeah, it took a bit time what took some solutions what you built to get out this feeling.

718
00:38:19,980 --> 00:38:25,980
And what role will open source play in enterprise AI?

719
00:38:25,980 --> 00:38:53,980
If you look at the benchmarks open source gets better and better but on the other hand, if you want to get to get the quality what this frontier models also provides you, you also need a lot of compute and I think when you calculate it, I'm not sure if if you want to really have the strong models and the really.

720
00:38:53,980 --> 00:39:17,980
And really amazing quality what they deliver in open source, I think there are a model is out in the field which are close to them, but to to pay this compute i'm not sure if you are cheaper compared to this pay as you go it always depends on your solution on your usage, a lot of different factors.

721
00:39:17,980 --> 00:39:42,980
But to run it on on a MacBook in this quality is not possible, this means you really really need this strong in the VTR to use and I think only to get them is is not as easy and also to run them also cost a lot of energy and yeah.

722
00:39:42,980 --> 00:39:48,980
I'm not sure if if this will be beneficial for companies.

723
00:39:48,980 --> 00:40:01,980
But also depends depends some companies has no other other possibilities because they have to run it on premise due to regulations and then it's a good possibilities.

724
00:40:01,980 --> 00:40:17,980
And also if it's capabilities without running through the internet, but I think in the most cases, yeah, you have to use to use the big models from providers in the cloud.

725
00:40:17,980 --> 00:40:32,980
When you get a chance to say what should published Microsoft tomorrow and found which feature to show the B.

726
00:40:32,980 --> 00:40:45,980
What I see is the the strong trend to get within real agent orchestration agent organization.

727
00:40:45,980 --> 00:41:00,980
This means you have daily in the end how it currently works today of see all your have leads your team players and you can directly talk with team players you can talk with the management they can delegate us.

728
00:41:00,980 --> 00:41:28,980
My point of view that's the way how you can get out a lot of strongness out of it this means to have different experts strong experts you have some more generalist and they can figure out who is the right right person to do this task for sure you already can work with agents and sub agents but the orchestration for a real strong organization is a bit missing.

729
00:41:28,980 --> 00:41:52,980
And I think that's the secret source to get a good quality when you want to build a multi powerful solution and that's the end what I built the last months by my own to try for platform who works on this way and it would be nice if we also see this first party from Microsoft.

730
00:41:52,980 --> 00:42:06,980
That's really cool and when we think about the future did you think we still need pro coders or a little something.

731
00:42:06,980 --> 00:42:09,980
This is a drop that will be gone.

732
00:42:09,980 --> 00:42:38,980
For me it's not a job which will be gone I think it will be more important but with the right mindset and that's I think in a lot of areas the point you need to write mindset you need to adapt this and then you have two kinds of persons you have a person who survive in his job and you have persons who don't adapt to don't have the right mindset to work with AI.

733
00:42:38,980 --> 00:43:02,980
They have some hard lives I think developers are good because you need the ideas you need the creativity you need the business logic and you have to tell the AI what they have to do you have to understand it also I think it's a question of months where AI will over paste the knowledge of humans.

734
00:43:02,980 --> 00:43:31,980
That's I think not not in secret but also you need someone who is accountable who can also make sure that everything what the AI produce is also in the interest of the company and I think for this you need senior developers you need a person who really adapt the technologies and then you can 100 extra productivity.

735
00:43:31,980 --> 00:44:00,980
This is also what I see here I'm able to work on multiple projects and parallel and yeah I would really say I 100 x my productivity and that's important that people see this possibilities and use them and march to say it's shit what the AI produce it's insecure because that's not a right mindset what you need for the future.

736
00:44:00,980 --> 00:44:14,980
And can you give some advice to people they are all overwhelmed by how fast this AI stuff is moving.

737
00:44:14,980 --> 00:44:35,980
Yeah don't get overwhelmed in the end everyone has to start on some on some point and it's important that you're curious that you try out that you build your own agent that you build your own tools whatever.

738
00:44:35,980 --> 00:45:04,980
And try a small clots subscription or code X whatever and try to to build your application deploy a model in foundry I already explained it is very easy and try to build your intelligent app and as I said sometimes it's only one prompt you get exactly the result what you want to have for sure it's not production production ready but in the end you learn a lot when you build such an app and you see also how strong.

739
00:45:04,980 --> 00:45:31,980
And what I also saw in the past you buy a lot of tools to make you a bit more productive or to solve a small challenge today on if I have to do some manual work or whatever I quickly caught me a solution and in the most cases I spent less time to produce the app which I would pay a lot of money in the past.

740
00:45:31,980 --> 00:45:51,980
And then I compared to the time which I would waste when I do it manually and the end it's a bit scarier if you see how lazy you get and if you have to do all this work manually again I think that's impossible.

741
00:45:51,980 --> 00:46:09,980
And also you have to manage a bit also your expectations because when you adapt AI everywhere you have such a high productivity this means also your expectation to yourself as such high that you deliver so much output over the day.

742
00:46:09,980 --> 00:46:24,980
And then you have a bit more manual work or stuff like this and your output is not on this level you're a bit annoyed and that's also something where you have to set your own expectations.

743
00:46:24,980 --> 00:46:33,980
When you could build an AI product with unlimited resource of what would it be.

744
00:46:33,980 --> 00:46:50,980
Oh, that's a good question. In the end I think when I can really dream and can build something with unlimited resources I would rethink the way how operating system works.

745
00:46:50,980 --> 00:47:02,980
And what we have today with windows make Linux whatever it's also it's everything is designed for you someone clicks on the mouse and do something.

746
00:47:02,980 --> 00:47:25,980
If I would build something by my own I would put AI everywhere where you can I would yeah design it on a way that it's AI native and this means that everywhere where it's possible AI can took over my work and I only have to think and make hey do this and I can can do it.

747
00:47:25,980 --> 00:47:35,980
When you have to say there's one AI bus word there you are tired of which one is it.

748
00:47:35,980 --> 00:47:38,980
Hey, I itself.

749
00:47:38,980 --> 00:47:51,980
Yeah, and you you read it everywhere you read it on a lot of products you read it also on a lot of features where no AI is in for example in Microsoft in tune Microsoft published and

750
00:47:51,980 --> 00:48:08,980
off-bodding agent or yeah, an off-bodding agent to retire machine which are non-active since 90 days and that's for me and linear workflow if a machine don't connect for 90 days retired.

751
00:48:08,980 --> 00:48:37,980
And then I've been named it agent and that's a bit the whole marketing thing around AI and agents and I'm tired of all these marketing things for stuff where AI is put on where I don't is not needed because there's still a lot of valid cases where standard workflow works because you you build it once you test it and when it works it works.

752
00:48:37,980 --> 00:48:58,980
And you have a kind of yeah, very answers how they act it's not always thousand percent sure that it always behave on the same way and as I said there are a lot of valid cases for classical flows or implementations and not everywhere.

753
00:48:58,980 --> 00:49:06,980
You have to put a ion to sell the product but that's what a lot of vendors do.

754
00:49:06,980 --> 00:49:18,980
You have out of the book modern end point and I key driven IT management is there not look you can say people should read.

755
00:49:18,980 --> 00:49:40,980
Yes, but in the end what I like is the one percent of I forgot the name of the book but what I really encourage people is more or less to read people about mindset because mindset is very important to.

756
00:49:40,980 --> 00:50:05,980
Yeah, be successful in a lot of things in in in your life but also in thinking AI on the right way and as I said technical the technical challenges are not as high as they was maybe one or two years ago for me it's more to have the right thinking to have the right mindset and this books which which I can.

757
00:50:05,980 --> 00:50:34,980
Recommend and what what can people learn from your book my book in the end I'm MVP for foundry and in tune to different worlds and I try to combine both words to show how I can help in the world of modern end point management how you can build your own solutions how you can build your own core pilots I wrote this book.

758
00:50:34,980 --> 00:51:03,980
I publish it in January one year ago but in the meantime yeah a lot of things are already outdated to do to the speed but in the end to inspire a bit to get my thinking I think this book will help to also understand the base technologies to implement AI I think that's everything in this book but if you publish today a book about AI and the life cycle is.

759
00:51:03,980 --> 00:51:21,980
Yeah very very short because as I said the the pace is too high yeah I do the m 365 content but feel free to tell what's your favorite conference to speak it.

760
00:51:21,980 --> 00:51:45,980
My favorite conference was work place ninja and Danemark I think it was around thousand people who participate here and I'm on a lot of events but that's from the outstanding although organization is how the venue is how the

761
00:51:45,980 --> 00:52:12,980
collaboration sorry experts life Danemark how the collaboration of all the participants and the atmosphere it was amazing because I am often on conferences which focus on one single topic and this was really bright topic but it was was one community and it was a very nice event.

762
00:52:12,980 --> 00:52:21,980
Awesome and what was the last thing you learned that blow your mind.

763
00:52:21,980 --> 00:52:45,980
And the goal feature from codex and now also on cloud you said a goal and I think in the meantime it runs three days three days without break you said a goal let the agent fulfill this goal goal and he will first end when this goal is really really

764
00:52:45,980 --> 00:53:14,980
really really a soft and that's really mind blowing because in the past you have made this session from one hour you say hey implement me this feature it's implement this you tested it doesn't work you make another try and yeah after some iterations you get it on a level where you want to have it but now you said slash goal and you say implement me the application which is this and this feature and some days later

765
00:53:14,980 --> 00:53:19,980
you have it perfectly working and that's really mind blowing.

766
00:53:19,980 --> 00:53:23,980
It's also correct itself.

767
00:53:23,980 --> 00:53:38,980
Yeah in the end you said a goal and it's always correct itself until this goal is really reached and when you ride in I want to have 100% working application

768
00:53:38,980 --> 00:53:55,980
then it makes a lot of tests and everything and yeah as I said it runs in the meantime three days in a row I'm curious when when it ends but that's really mind blowing from you.

769
00:53:55,980 --> 00:54:14,980
Yeah we run in a little bit out of time so then my last question is when people look back at this AI area 10 years from now what do you hope your contribution to the industry will yeah will have been.

770
00:54:14,980 --> 00:54:35,980
For me for everything what I do in a community for me it's important to inspire to show how you can build a solution based on the technologies what you have for me it's also important to help the people to get the right thing into right mindset and I often

771
00:54:35,980 --> 00:54:57,980
have two or three steps before the technologies this means I try also to think a lot in the future and all already start today to think about how this could be beneficial for us or for the community or for customers whatever and also to share as much as I can.

772
00:54:57,980 --> 00:55:07,980
This means that's really what gives me energy to share knowledge to share ideas and bring it to the community.

773
00:55:07,980 --> 00:55:23,980
I say thank you and the last words I give you to the audience and one thing I only mentioned is yeah all listeners you find all information about Yannick in the show notes so yeah what's your last words.

774
00:55:23,980 --> 00:55:50,980
First of all thank you very much to invite me was a lot of fun to be part of this podcast and my really last words is don't think about really try it out start to deploy your model try it out build your own solutions because technology is moving fast and you should jump on the train and to yeah educate yourself.

775
00:55:50,980 --> 00:55:57,980
To understand how air I works and how you can bring benefit to your enterprise.

776
00:55:57,980 --> 00:56:02,980
Yeah then thank you Yannick this was amazing bye and have a great day.

777
00:56:02,980 --> 00:56:04,980
Bye.

778
00:56:04,980 --> 00:56:15,040
[BLANK_AUDIO]