WEBVTT

1
00:00:00.320 --> 00:00:06.280
Every industry industry. Inside Analysis is
your source of information and insight about how

2
00:00:06.280 --> 00:00:10.919
to make the most of this exciting
new era. Learn more and Inside analysis

3
00:00:11.000 --> 00:00:16.000
dot com, Insideanalysis dot com and
now here's your host, through Eric Kavanaugh.

4
00:00:20.719 --> 00:00:23.719
Ladies and gentlemen, Hello, and
welcome back once again to the only

5
00:00:23.800 --> 00:00:27.960
coast to coast radio show in the
USFA that's all about the information economy.

6
00:00:28.000 --> 00:00:32.000
It's time for Inside Analysis. You're
truly Eric Kavanaugh here, and folks,

7
00:00:32.039 --> 00:00:36.399
I am very excited to have frankly
a legend in the industry with us.

8
00:00:36.399 --> 00:00:39.719
Today we're going to be talking with
Michael Bertholdt. He is the founder and

9
00:00:39.799 --> 00:00:45.119
CEO of a company called NIME that's
spelled k n i M and they are

10
00:00:45.159 --> 00:00:50.600
an open source analytics and data science
platform, a visual platform for doing data

11
00:00:50.679 --> 00:00:53.560
science, which is good stuff because
even though there are lots of people who

12
00:00:53.560 --> 00:00:57.799
can code very well, almost anyone
can look at visuals to move boxes around

13
00:00:57.840 --> 00:01:00.520
on a screen. So that's what
they figured out. So at that,

14
00:01:00.600 --> 00:01:03.759
Michael, welcome to Inside Analysis.
Tell us a bit about yourself and NIME

15
00:01:03.840 --> 00:01:07.120
in which you folks are working on
these days. Thanks thanks for the invite,

16
00:01:07.280 --> 00:01:11.719
Eric and having us run the show. As you already said, NIME

17
00:01:11.840 --> 00:01:15.120
is very much about visual programming low
code for data science. And we started

18
00:01:15.120 --> 00:01:19.560
that many years ago as a platform
really for as a workbench almost at my

19
00:01:19.840 --> 00:01:23.359
group at the University of University of
Constants, to be able to kind of

20
00:01:23.439 --> 00:01:27.480
deploy our research results to the real
world to practitioners wanting to use that.

21
00:01:29.040 --> 00:01:34.519
And it's grown from there to become
one of the only open source visual data

22
00:01:34.640 --> 00:01:38.560
flow platforms for doing anything you want
to do with data. And I mean

23
00:01:38.640 --> 00:01:42.560
I kind of I'm always a little
bit careful about calling it data science because

24
00:01:42.560 --> 00:01:47.439
that often scares people because they say, I just want to do data wrangling.

25
00:01:47.480 --> 00:01:49.120
I don't care about the science seed
beads, and that's kind of scares

26
00:01:49.159 --> 00:01:53.200
them off. That's what nine does
is want so a lot of applications that

27
00:01:53.200 --> 00:01:57.799
we see in real life is just
in large airports, just getting data in

28
00:01:57.840 --> 00:02:01.359
the right shapes from many different sources. Yeah, and I will tell you

29
00:02:01.400 --> 00:02:05.879
I'm a huge fan of open source. In fact, we built a website

30
00:02:05.920 --> 00:02:07.960
a number of years ago and we
feed it with a technology that we built

31
00:02:07.960 --> 00:02:13.800
called media lens. It's called inside
open source. And the reason we launched

32
00:02:13.800 --> 00:02:16.759
it was because I realized there is
so much happening in the open source world.

33
00:02:16.800 --> 00:02:20.800
There's the Apache Foundation, as the
Linux Foundation. There are lots of

34
00:02:20.800 --> 00:02:25.479
other projects outside of those organizations as
well, But in the origin of open

35
00:02:25.560 --> 00:02:30.199
source I found fascinating. So I
first started researching this in two thousand and

36
00:02:30.240 --> 00:02:35.960
five when I was working for the
Data Warehousing Institute as their web evangelist,

37
00:02:36.400 --> 00:02:39.439
and Katrina had just struck New Orleans. I just moved out of New Orleans,

38
00:02:39.479 --> 00:02:43.240
so we watched it all happen on
TV. I was just, of

39
00:02:43.280 --> 00:02:47.520
course terrifying. But I remember that
the senators from Louisiana asked for a quarter

40
00:02:47.599 --> 00:02:53.759
of a trillion dollars to rebuild southern
Louisiana. And I happened to know through

41
00:02:53.759 --> 00:02:57.520
a past life and past clients in
the government space down there, that the

42
00:02:57.599 --> 00:03:00.479
politicians are good at making money disappear. And I thought to myself, this

43
00:03:00.560 --> 00:03:04.639
is a very bad situation, because
if all that money just floods in,

44
00:03:04.680 --> 00:03:07.919
it's going to flood right out,
not where anyone intended, and a lot

45
00:03:07.960 --> 00:03:09.680
of it is just going to disappear. So I went on this high horse,

46
00:03:09.719 --> 00:03:14.599
if you will, and started doing
research, and it took me to

47
00:03:14.800 --> 00:03:19.240
open source, and I put forth
this theory about open source government, about

48
00:03:19.240 --> 00:03:23.360
publishing all the government data such the
citizens can see where the money goes and

49
00:03:23.439 --> 00:03:27.919
understand. And I basically said,
look, with the sarbainez Ox the Act

50
00:03:27.960 --> 00:03:30.800
in the States, which came out
of the Enron tobacle, corporations had to

51
00:03:30.960 --> 00:03:35.080
document their processes for how they come
up with their numbers and they had to

52
00:03:35.159 --> 00:03:38.159
be very transparent about that stuff.
And I thought, well, if corporate

53
00:03:38.120 --> 00:03:39.919
America has to do it, why
doesn't the government do it as well?

54
00:03:40.280 --> 00:03:45.599
And people thought I was crazy,
But a couple things happened that were amazing.

55
00:03:45.960 --> 00:03:49.240
One guy paid attention. Out of
forty thousand people I emailed. One

56
00:03:49.280 --> 00:03:52.919
guy paid attention, and he worked
for the Heritage Foundation. He went and

57
00:03:53.000 --> 00:03:57.719
talked to He basically testified to Congress
and said, we can have citizen auditors.

58
00:03:57.719 --> 00:04:00.439
This stuff really can happen. And
he really leveraged the impt A tour

59
00:04:00.520 --> 00:04:03.639
of TDWY and sure enough they did
it. House passed the bill, Senate

60
00:04:03.680 --> 00:04:06.879
passed the bill, co sponsored by
a guy named Barack Obama who was a

61
00:04:06.879 --> 00:04:11.080
Senator for Illinois, and then President
George W. Bush, believe it or

62
00:04:11.120 --> 00:04:15.800
not, signed the federal Funding Accountability
and Transparency Act in September six, twenty

63
00:04:15.840 --> 00:04:18.079
six, two thousand and six,
and I almost fell off my chair.

64
00:04:18.120 --> 00:04:21.399
I was like, Wow, they
actually did it. But the reason I

65
00:04:21.519 --> 00:04:26.040
bring this up is to talk about
the power of open source. And I

66
00:04:26.120 --> 00:04:31.759
realized at the time the Apache web
server had just surpassed the Microsoft Web server

67
00:04:31.920 --> 00:04:34.399
as the number one web server.
And I thought to myself, well,

68
00:04:34.439 --> 00:04:39.639
that's very interesting. And I was
also studying service oriented architecture at the time,

69
00:04:39.639 --> 00:04:42.800
and I thought, well, if
you have all this open source code

70
00:04:42.959 --> 00:04:46.240
and you have a service oriented architecture, you should be able to plug and

71
00:04:46.279 --> 00:04:48.160
play and sort of take stuff out
and put stuff in and have a very

72
00:04:48.160 --> 00:04:51.759
composable environment. And I thought,
well, that's not going to be very

73
00:04:51.759 --> 00:04:56.759
good news for the SAPs and the
Oracles of the world, because they liked

74
00:04:56.800 --> 00:04:59.959
the monolith. They like control,
and they have control of all that stuff.

75
00:05:00.079 --> 00:05:03.040
And it took longer than I thought, but about ten years later open

76
00:05:03.079 --> 00:05:09.519
source just blew up the market with
Haddoub and with the Kafka and of course

77
00:05:09.600 --> 00:05:13.959
you have nine. So tell me
a bit about the open source foundation of

78
00:05:14.120 --> 00:05:17.319
NIME and what drove your decision there
and what it means for your customers.

79
00:05:19.920 --> 00:05:25.000
That's a very interesting story. I
didn't know about that open data movement coming

80
00:05:25.040 --> 00:05:30.000
from Katrina break in the old d
is so NIME is a bit atypical in

81
00:05:30.120 --> 00:05:33.000
its open source models. I mean
fundamentally, there are really three different spays

82
00:05:33.079 --> 00:05:36.439
to do that. You can have
distributions like Linox, and you're essentially making

83
00:05:36.519 --> 00:05:42.040
money by packaging them up nicely and
supporting them, but essentially there's not really

84
00:05:42.480 --> 00:05:44.959
new code that you're adding around it. Not quite true. I mean they

85
00:05:45.000 --> 00:05:47.560
are little installers and that type of
stuff, but that's fundamentally the idea behind

86
00:05:47.600 --> 00:05:51.959
distributions. You can also have what
people often refer to as open corea where

87
00:05:51.959 --> 00:05:56.079
you have something that you open source, but it's kind of it's more of

88
00:05:56.120 --> 00:05:58.800
a teaser, it's more baitware,
and if you really want to use this

89
00:05:58.879 --> 00:06:01.639
introduction, you have to buy some
commercial bits and pieces around it. And

90
00:06:01.680 --> 00:06:05.160
then you have, of course also
the longo debs that are essentially databases.

91
00:06:05.360 --> 00:06:09.240
That type of stuff also maybe even
a data breaks with Spark. They have

92
00:06:09.399 --> 00:06:13.120
really cool open source technology and they
make it accessible to you in the cloud

93
00:06:13.879 --> 00:06:16.720
for a charge. Time is different
in that we have one open source piece

94
00:06:16.800 --> 00:06:23.120
that's the analytics platform that allows anybody
who wants to build these workflows and execute

95
00:06:23.120 --> 00:06:26.920
these workflows and pretty much do anything
they want to do with data. And

96
00:06:27.000 --> 00:06:30.240
then we have a commercial complement software
to that, which we call the nine

97
00:06:30.319 --> 00:06:34.959
Business Hub, which allows people to
productionize that and collaborate. So when you

98
00:06:35.120 --> 00:06:40.040
have more than one person using the
analytics platform in your organization and you want

99
00:06:40.079 --> 00:06:43.399
to deploy that as a web interface
or a west service, or you want

100
00:06:43.439 --> 00:06:46.800
to collaborate and have compliance and governments
kind of features, that's when you buy

101
00:06:47.199 --> 00:06:53.120
the hub from us. And the
reason the open source platform is open source.

102
00:06:53.680 --> 00:06:58.199
The open source analytics platform is open
source is I'm not too religious around

103
00:06:58.240 --> 00:07:00.240
it, but to me, it's
in the data sigen and field in particular,

104
00:07:00.759 --> 00:07:05.680
you can't exist with proprietary software.
There's so much cool new stuff going

105
00:07:05.800 --> 00:07:11.319
on on a daily basis in resourched
groups and other types of environments that you're

106
00:07:11.399 --> 00:07:15.360
essentially standing on the shoulder of many, many giants. A lot of functionality

107
00:07:15.480 --> 00:07:18.879
inside NIME is actually based on open
source libraries, so it seems kind of

108
00:07:19.560 --> 00:07:25.839
unfair almost to put that into a
proprietary umbrella. And also it enables us

109
00:07:25.920 --> 00:07:30.279
to be fair to get in in
roles, make inrolls into academic into teaching

110
00:07:30.360 --> 00:07:32.920
environments much easier because they can just
use the open source platform for teaching.

111
00:07:33.279 --> 00:07:40.040
But it also we have of open
source contributors that are contributing additional functionality into

112
00:07:40.079 --> 00:07:43.480
the line platform. So it's really
I'd like to see it as a win

113
00:07:43.560 --> 00:07:47.839
win situation also for our customers because
they essentially get a lot of maintained functionality

114
00:07:47.959 --> 00:07:53.160
from us. In addition, they
have access to all of those community functions

115
00:07:53.240 --> 00:07:56.439
out there. Yeah, it's very
cool. I mean, there are a

116
00:07:56.439 --> 00:07:59.199
lot of good things about open source. One of the things that I've heard

117
00:07:59.199 --> 00:08:03.319
over the years is that bad code
goes away because all these eyes can see.

118
00:08:03.439 --> 00:08:07.800
Now. The one shortcoming up come
across is that the open source project

119
00:08:09.160 --> 00:08:13.560
gets to MVP status if you will, minimum viable product, and then doesn't

120
00:08:13.560 --> 00:08:16.120
typically go past that just because it
works. Now we just kind of move

121
00:08:16.240 --> 00:08:20.600
beyond that. But what are your
thoughts about about that, in particular about

122
00:08:22.279 --> 00:08:26.680
how you make sure that you have
truly finished products and that you're able to

123
00:08:26.279 --> 00:08:31.960
deliver robust platform analytics ongoing for all
of your clients. How much effort does

124
00:08:33.000 --> 00:08:37.320
that take internally from your developers to
stay on top of the platform and make

125
00:08:37.360 --> 00:08:41.320
sure everything's working. That's a very
good compoint. I mean, I tend

126
00:08:41.360 --> 00:08:46.320
to joke that ninety nine percent of
all PhD projects turn into open source projects

127
00:08:46.360 --> 00:08:48.320
and then they kind of die away
and feel away and never really turn into

128
00:08:48.399 --> 00:08:54.159
something useful in production. Now,
I'm probably about half of our development team,

129
00:08:54.559 --> 00:09:00.240
like Ady, developers at nine are
focused exclusively on the analytics platform and

130
00:09:00.360 --> 00:09:07.879
just making sure that Core works and
works in a professional environment. We have

131
00:09:07.240 --> 00:09:11.879
our own extensions, which are of
course maintained by ourselves, so we have

132
00:09:11.000 --> 00:09:16.519
the same quality assurance there. We
have what we call trusted Community extensions,

133
00:09:16.559 --> 00:09:20.399
where we're in close collaboration with the
community contributing those extensions, so we can

134
00:09:20.440 --> 00:09:24.360
also make sure there's quality insurance there
as well. And then there is i'd

135
00:09:24.440 --> 00:09:28.320
call it the long tail of extensions
that are experimental, right. The nice

136
00:09:28.360 --> 00:09:33.639
thing is that everybody can use those
and play with those and explore new technologies,

137
00:09:33.120 --> 00:09:37.360
and then when we see increased usage
of some of these experimental extensions,

138
00:09:37.440 --> 00:09:41.759
we can move them into the trusted
extensions as well. Interesting that makes a

139
00:09:41.799 --> 00:09:46.240
lot of sense. So you're trying
out things, you've got these extensions in

140
00:09:46.360 --> 00:09:50.919
trials basically, and then once you
see there's a lot of activity, then

141
00:09:50.960 --> 00:09:54.279
you grow some developer support behind it. To harden it is the term we

142
00:09:54.360 --> 00:09:56.840
typically use right to make it sure
that it's bulletproof, that it really does

143
00:09:56.919 --> 00:10:01.080
what you want it to do.
E makes a lot of sense, and

144
00:10:01.440 --> 00:10:05.879
you do end to end. NIME
does everything from data ingestion, data pipelines,

145
00:10:05.480 --> 00:10:09.600
number crunching, model building, all
that kind of fun stuff is in

146
00:10:09.679 --> 00:10:11.600
the nine analytics platform. Is that
right? Yes, that's true. So

147
00:10:11.679 --> 00:10:16.840
we have everything from about the ETL
part loading the data. We can access

148
00:10:16.960 --> 00:10:20.720
about four hundred different data sources.
You can access databases, strange file formats

149
00:10:20.879 --> 00:10:26.000
x so of course we can also
execute bits and pieces on different execution environments

150
00:10:26.120 --> 00:10:31.440
like doing the ETL directly insider database
or in snowflake or in data breaks or

151
00:10:31.480 --> 00:10:33.679
in our loop in the old days. And then we go all the way

152
00:10:33.799 --> 00:10:37.840
to visualization. The analytics functionality,
and a lot of that, as I

153
00:10:37.919 --> 00:10:41.440
said before, is of course based
on each charts for the visualization a lot

154
00:10:41.480 --> 00:10:46.840
of pythen libraries see libraries, our
libraries, Java libraries. For some of

155
00:10:46.879 --> 00:10:50.039
the machine learning functionality, we have
integrations with TensorFlow if you wanted to do

156
00:10:50.159 --> 00:10:54.960
that. We have integrations with the
other deep learning libraries, we have integrations

157
00:10:54.080 --> 00:10:58.879
with xg boosts. Pretty much everything
is in there, but the other pieces.

158
00:11:00.080 --> 00:11:03.120
So often when people talk about data
science stainly mean this kind of from

159
00:11:03.200 --> 00:11:07.840
the data to the reporter, to
the endpoint or to the model. But

160
00:11:07.159 --> 00:11:11.879
the business hub then also covers the
rest of this journey, right, deploying

161
00:11:11.960 --> 00:11:16.039
it to others, managing it,
three training models when needed, and monitoring

162
00:11:16.120 --> 00:11:20.720
their their performance well. Right,
because at the end of the day,

163
00:11:20.000 --> 00:11:24.360
you want these algorithms to connect into
your business, whether it's spear of marketing

164
00:11:24.679 --> 00:11:30.639
or for manufacturing or supply chain or
whatever it is. You want it to

165
00:11:30.879 --> 00:11:35.600
affect some outcome in the business,
and so that involves connecting to operational systems,

166
00:11:35.679 --> 00:11:39.879
right. It involves connecting to EERP
systems or CRM systems or things of

167
00:11:39.919 --> 00:11:43.679
this nature. That's where the magic
happens. And a lot of times that's

168
00:11:43.759 --> 00:11:48.080
the hard part, right. I
Mean, I've heard many stories about models

169
00:11:48.159 --> 00:11:52.840
that just don't get deployed because maybe
the companies didn't have the wherewithal or they

170
00:11:52.919 --> 00:11:56.240
didn't have the expertise to do it
properly. But being able to plug the

171
00:11:56.320 --> 00:12:03.320
algorithms into operational systems and then monitor
how those models perform and switch them out

172
00:12:03.399 --> 00:12:07.360
right because you've got your production model
and have your challenger models that are sitting

173
00:12:07.399 --> 00:12:11.159
at by the wayside waiting to get
pulled in and being able to switch over

174
00:12:11.799 --> 00:12:16.080
to a new model when a model
that's in production starts faltering. That's a

175
00:12:16.200 --> 00:12:18.600
critical piece and that, I guess, is that done in business hub with

176
00:12:18.679 --> 00:12:22.559
you folks. That's something on the
business subside exactly. So as business hup,

177
00:12:22.639 --> 00:12:28.080
you can deploy models which really are
deployed nine workflows. You can deploy

178
00:12:28.120 --> 00:12:31.840
in them as a rest service or
as a bad application, and then people

179
00:12:31.919 --> 00:12:35.639
consume it, and you can constantly
monitor what's happening in production and then potentially

180
00:12:35.759 --> 00:12:39.279
replace we train, or just alert
the data science team and just say,

181
00:12:39.320 --> 00:12:43.000
hey, this is so out of
backwith reality. We don't really know how

182
00:12:43.039 --> 00:12:45.879
to fix that, do something about
it. The nice thing is that you

183
00:12:45.960 --> 00:12:48.799
don't have to switch code in between. Right in the old days, we

184
00:12:48.960 --> 00:12:52.559
always somebody coded the model of the
strains in some strange language and then it

185
00:12:52.720 --> 00:12:58.440
was reprogrammed by it into some production
language. On our case, on the

186
00:12:58.519 --> 00:13:01.840
hubside, the workflow that was trained
is also the one that runs in production.

187
00:13:03.279 --> 00:13:07.480
That's interesting. So one of the
other hurdles that people are running into

188
00:13:07.720 --> 00:13:11.960
is when they use Jupiter notebooks to
write their model or to build their model

189
00:13:11.960 --> 00:13:13.600
to test with data, and then
they want to go put that into production,

190
00:13:15.200 --> 00:13:18.320
and it's just this step by step
tedious process of copying over code and

191
00:13:18.519 --> 00:13:22.399
values and all these things, and
that falters often. That, from what

192
00:13:22.480 --> 00:13:26.639
I understand, is a really serious
problem. But I guess do you not

193
00:13:26.840 --> 00:13:30.840
have that challenge because you're not using
the Jupiter notebooks typically, and people are

194
00:13:30.960 --> 00:13:35.480
just in the environment in the analytics
platform building out their models after they pull

195
00:13:35.519 --> 00:13:39.919
in their data, et cetera.
So you're already production ready when the process

196
00:13:39.960 --> 00:13:45.080
begins. Is that about right?
That's a very nice summary. Yes,

197
00:13:45.240 --> 00:13:48.480
So what you use, what you
use on the creation site when you actually

198
00:13:48.519 --> 00:13:52.960
train the model is exactly that piece
of the workflow gets then moved into production

199
00:13:52.039 --> 00:13:56.000
and executed by exactly the same engine, so you don't also have a translation

200
00:13:56.120 --> 00:14:01.240
issue there. The other piece that
people often lose in this going from training

201
00:14:01.519 --> 00:14:05.919
to production is all of the feature
engineering that you did, all the feature

202
00:14:05.960 --> 00:14:09.759
transformations that you tend to lose them. So you only can take the model

203
00:14:09.759 --> 00:14:13.559
and move that in production, but
you can't do the transformations. And in

204
00:14:13.679 --> 00:14:16.919
Lime you can grab automatically the part
of the workflow that has the transformations and

205
00:14:18.039 --> 00:14:22.000
the model applying the model, take
that rep that automatically and deploy it to

206
00:14:22.200 --> 00:14:28.480
line business up. Well, that's
pretty cool. And you cover also two

207
00:14:28.519 --> 00:14:33.360
different industries, right, so you
do insurance healthcare? Would imagine financial services

208
00:14:33.519 --> 00:14:37.919
all sorts of different industries because it's
more of a horizontal solution instead about right,

209
00:14:39.559 --> 00:14:41.639
Yes, that's absolutely true. We
have customers and users in pretty much

210
00:14:41.840 --> 00:14:46.240
every industry. Yeah. Well,
and we're going to talk about large language

211
00:14:46.240 --> 00:14:50.480
models here in a minute in our
next segment, but before we get there,

212
00:14:50.559 --> 00:14:52.120
I'll just throw out one of my
theories to you and see what you

213
00:14:52.159 --> 00:14:58.960
think about this. To me,
this explosion of AI through foundational model,

214
00:15:00.120 --> 00:15:05.000
including large language models, is really
a major call to action for organizations to

215
00:15:05.080 --> 00:15:09.120
get their data house in order.
And what I mean is that data governance.

216
00:15:09.159 --> 00:15:11.399
If you don't have data governance,
if you don't even know what data

217
00:15:11.399 --> 00:15:16.360
governance is from an organizational perspective,
you're going to have a hard time responsibly

218
00:15:16.480 --> 00:15:20.600
leveraging AI. Would you agree that
companies really do need to take a very

219
00:15:20.720 --> 00:15:26.440
hard look at their end to end
data management life cycles processes, understand governance,

220
00:15:26.480 --> 00:15:31.360
who gets access to what data?
Even understanding a broad inventory of your

221
00:15:31.440 --> 00:15:35.879
data sources. Would you agree that
is paramount to do that before pulling the

222
00:15:35.960 --> 00:15:41.039
trigger on some AI. I totally
agree. And the funny in the way,

223
00:15:41.120 --> 00:15:46.000
it's funny that we've been preaching this
for also data science processes for a

224
00:15:46.039 --> 00:15:48.840
long time, this government's topic,
and nobody really cared, and now people

225
00:15:48.919 --> 00:15:54.559
really care by breaking up. Isn't
that fascinating? I mean, I just

226
00:15:54.799 --> 00:15:58.200
I've been in this business a long
time. I've been talking about data governance,

227
00:15:58.240 --> 00:16:00.720
analytics, AI, all this stuff
for twenty years, right, And

228
00:16:02.200 --> 00:16:04.080
we talked about data governance twenty years
ago and fifteen years ago and ten years

229
00:16:04.080 --> 00:16:07.120
ago, and basically nobody was doing
it. I mean, you couldn't even

230
00:16:07.279 --> 00:16:11.759
it wasn't even easy to do because
you could either control access at the database

231
00:16:11.879 --> 00:16:15.200
level, which is hard to access
controls, or at the application level.

232
00:16:15.279 --> 00:16:18.399
But there's nothing in the middle.
And really it's in the middle. And

233
00:16:18.519 --> 00:16:22.799
now with the cloud, that's one
of the nice things about the cloud is

234
00:16:22.840 --> 00:16:26.639
that it is this de facto marshaling
area for functionality and data. And now

235
00:16:26.759 --> 00:16:32.399
we have the capacity to apply very
fine grain controls on things, on data

236
00:16:32.480 --> 00:16:37.080
sets, on types of data.
For example, we can scan and find

237
00:16:37.159 --> 00:16:41.039
PII and then know, okay,
flag this as sensitive. There are lots

238
00:16:41.039 --> 00:16:44.200
of things we can do these days
that we just kind of couldn't do ten

239
00:16:44.320 --> 00:16:45.240
years ago. Real quick, one
minute, what do you think is that

240
00:16:45.279 --> 00:16:48.320
about? Writer? We finally can
do this stuff. Until we are doing

241
00:16:48.399 --> 00:16:53.639
it, what do you think.
I'd probably explained it slightly differently and say

242
00:16:53.720 --> 00:16:57.360
we could have done it probably before
as well, at least some of those

243
00:16:57.440 --> 00:17:02.320
aspects, but people just did didn't
care enough because there was not enough arm

244
00:17:02.399 --> 00:17:06.240
in it. Now But now that
everybody who does anything with JENNYI is the

245
00:17:06.319 --> 00:17:11.599
danger of sending data anywhere people are
really really baking up and seeing the pain

246
00:17:11.680 --> 00:17:14.240
there. That's right. Well,
it is like I say, it's a

247
00:17:14.319 --> 00:17:18.359
call to arms, it's a call
to action that your organizations have got to

248
00:17:18.440 --> 00:17:21.960
do it because you don't want to
wind up in the crosshairs of an audit.

249
00:17:22.079 --> 00:17:22.799
You don't want to wind up with
a breach, you don't want to

250
00:17:22.839 --> 00:17:27.599
wind up getting sued by someone because
their information has now been leaked to sensitive

251
00:17:29.359 --> 00:17:30.680
the sensitive resources out there. Well, folks, don't touch that doll.

252
00:17:30.759 --> 00:17:36.160
We're talking all about AI and analytics
platforms, and next up we're going to

253
00:17:36.240 --> 00:17:38.880
dive into these large language models that
are just taking the business world by absolute

254
00:17:40.000 --> 00:17:42.200
storm. It's really quite fascinating to
watch. But don't touch out. That

255
00:17:42.319 --> 00:17:52.279
will be right back. You're listening
to Inside Analysis. Welcome back to Inside

256
00:17:52.319 --> 00:18:00.200
Analysis. Here's your host, Eric
Tabanac. All right, folks, back

257
00:18:00.200 --> 00:18:03.960
here on Inside Analysis talking to Michael
bertthol. He is the CEO and founder

258
00:18:03.000 --> 00:18:07.079
of a company called NIME. That's
k n I M Look these folks up

259
00:18:07.160 --> 00:18:11.720
online, an open source analytics platform. It's wonderful stuff. It's like a

260
00:18:11.799 --> 00:18:17.200
giant candy store for analysts to go
play and have fun. But I wanted

261
00:18:17.200 --> 00:18:19.640
to talk to you about these large
language models, Michael, and in particular,

262
00:18:19.839 --> 00:18:22.680
first of all, the open source
side of the equation. So Meta

263
00:18:23.000 --> 00:18:27.000
comes out with Lama and lama too. Open source. Open AI used to

264
00:18:27.039 --> 00:18:32.119
be open now it's not. Now
it's the ironically named open AI because it's

265
00:18:32.160 --> 00:18:37.400
a black box. And with the
technology this powerful, I believe we need

266
00:18:37.680 --> 00:18:41.480
we need open source. I don't
know that I would get behind a mandate

267
00:18:41.680 --> 00:18:45.960
that they must be open source,
but there needs to be some transparency into

268
00:18:47.119 --> 00:18:52.200
how these things are working, just
so that we can have our peaceful sleep

269
00:18:52.279 --> 00:18:56.960
at night to know that there are
bad actors involved somehow. I mean certainly

270
00:18:57.160 --> 00:19:02.440
for regulated industries like financial services.
If you bring it into some workflow for

271
00:19:02.599 --> 00:19:06.440
loan approval or something like that,
then you have to be able to explain

272
00:19:07.000 --> 00:19:08.960
how you came up to your answer. But what are your thoughts in general

273
00:19:08.960 --> 00:19:15.599
about open source versus closed source?
With these large language models, I think

274
00:19:15.680 --> 00:19:19.680
there's a lot of value in it. The problem is that, in my

275
00:19:19.799 --> 00:19:25.839
opinion, there's open sourcing large language
models isn't just about open sourcing the code,

276
00:19:26.440 --> 00:19:29.319
but you also need to open source
how it was actually trained. So

277
00:19:29.480 --> 00:19:33.359
in a sense you also need to
at least give open access to the data

278
00:19:33.440 --> 00:19:36.920
that was used for training. Because
even if I give you a model and

279
00:19:37.000 --> 00:19:41.119
it was trained on half copyrighted material
that it's going to spit out again when

280
00:19:41.160 --> 00:19:44.400
you use it, you wouldn't know
if you didn't have access to the training

281
00:19:44.480 --> 00:19:48.160
data, right that That's part of
that is is was it supposed to be

282
00:19:48.319 --> 00:19:53.559
used? And then I think the
other piece is that what some companies are

283
00:19:53.720 --> 00:20:00.359
open sourcing is only the code to
use the model for predictions later actually apply

284
00:20:00.440 --> 00:20:03.039
it, still don't know how it
was trained. So that's the third element

285
00:20:03.200 --> 00:20:10.039
that needs open sourcing. And then
I believe one of the key proprietary ingredients

286
00:20:10.079 --> 00:20:14.319
that a lot of these companies now
have is safeguarding code around it so that

287
00:20:14.480 --> 00:20:18.160
some types of answers don't get produced, some types of inputs aren't being accepted,

288
00:20:18.640 --> 00:20:22.880
and open sourcing that as well would
really really reveal their secret sauce.

289
00:20:22.960 --> 00:20:26.920
And I think that's why they are
the open eyes of the world are shying

290
00:20:26.920 --> 00:20:30.079
away from that one. Right,
No, it does make sense. I

291
00:20:30.160 --> 00:20:33.599
mean, we have proprietary code.
It's not new, but again, these

292
00:20:33.640 --> 00:20:37.039
are very very powerful engines. And
then there's another whole side of this equation,

293
00:20:37.160 --> 00:20:42.640
which is the RAG model Retrieval augmented
Generation, which upon great reflection,

294
00:20:42.920 --> 00:20:52.599
I believe will be the layer of
functionality for governance, for privacy, to

295
00:20:52.680 --> 00:20:56.160
a certain extent, for security,
for management. You know, a lot

296
00:20:56.200 --> 00:20:59.680
of that's going to get baked into
the RAG model, where you could bring

297
00:20:59.680 --> 00:21:02.519
example, Apple, before you hit
your prompt, before your prompt goes up,

298
00:21:02.559 --> 00:21:04.759
to the large language model. Have
a layer in between the checks and

299
00:21:04.839 --> 00:21:10.160
sees. Okay, and this is
already happening. Like I asked Gemini a

300
00:21:10.200 --> 00:21:15.480
couple of weeks ago how many electoral
votes are in Georgia and Arizona and some

301
00:21:15.599 --> 00:21:18.160
other state, and a thought for
a second. It said elections are complex

302
00:21:18.240 --> 00:21:23.119
and fast moving. We recommend you
use Google. It was a guard rail.

303
00:21:23.240 --> 00:21:26.039
That's a guardrail. They exactly built
that in to say no, no,

304
00:21:26.079 --> 00:21:29.559
no, no, we don't want
to touch that. Right, And

305
00:21:29.680 --> 00:21:32.599
that's in the RAG model. Right. That's not like trained in the model.

306
00:21:32.640 --> 00:21:36.880
That's outside the model, But it's
the workflow you have around the engine

307
00:21:37.240 --> 00:21:41.799
that's very very important. Right.
I totally agree with that, man.

308
00:21:41.880 --> 00:21:45.000
I mean that's what I call the
safeguards before. And I think sometimes it's

309
00:21:45.039 --> 00:21:48.200
probably not even part of the context
that's part of the RAG models, but

310
00:21:48.279 --> 00:21:51.680
it's really part of some safeguarding code
even around it. I mean we use

311
00:21:51.759 --> 00:21:55.839
that ATNAM as well. So we
have built in what we call KAY inside

312
00:21:55.880 --> 00:22:00.079
the analytics platform that allows you to
have a QA mode you ask questions do

313
00:22:00.400 --> 00:22:03.200
this and this, and excel out
does that look in a nine Doug fim

314
00:22:03.480 --> 00:22:06.400
and then it gives you X shows
you a couple of notes in nine,

315
00:22:06.799 --> 00:22:11.559
and we're of course filtering that these
notes do actually exist, because every now

316
00:22:11.559 --> 00:22:15.559
and then open AI, which we
use underneath the hood, hallucinates and invents

317
00:22:15.640 --> 00:22:18.319
notes that NIE probably should have but
we don't have it. That doesn't help

318
00:22:18.359 --> 00:22:22.960
the pro use. But that's a
very very simple way of code around the

319
00:22:23.799 --> 00:22:27.519
KAI that is just making sure that
what it spits out is reasonably useful.

320
00:22:29.200 --> 00:22:32.359
Mm hmm. Yeah, that's interesting, and you're going to see more and

321
00:22:32.480 --> 00:22:34.960
more of these AI agents. That's
what everyone is talking about now, are

322
00:22:36.000 --> 00:22:41.160
AI agents, which are like little
bots, semi autonomous bots that can do

323
00:22:41.359 --> 00:22:42.759
various things, and they can check
on each other and they can do all

324
00:22:42.880 --> 00:22:47.319
kinds of stuff. I mean,
it's very interesting to me when we talk

325
00:22:47.359 --> 00:22:49.400
about data science. We talked about
it before, it all seems to be

326
00:22:49.480 --> 00:22:53.880
getting subsumed now into AI. In
conversations about AI. Even though there are

327
00:22:53.960 --> 00:22:57.359
lots of different versions of AI,
right, I mean, there are traditional

328
00:22:57.400 --> 00:23:02.160
models, regression models like also of
old fashioned aif you well, it's still

329
00:23:02.279 --> 00:23:06.720
very powerful and still works, but
the new stuff is sucking all the auction

330
00:23:06.920 --> 00:23:10.240
out of the room. Isn't that
about right, Yeah, we see that

331
00:23:10.319 --> 00:23:11.559
as well, and sometimes I mean, I'm an old guy, but now

332
00:23:11.640 --> 00:23:15.680
I've seen this in the past.
Right when back propagation came along, everybody

333
00:23:15.799 --> 00:23:18.039
was suddenly using gradi into cent for
every problem. We just thought, hey,

334
00:23:18.119 --> 00:23:21.720
you can solve this directly, you
don't need to do Gradi intosst.

335
00:23:22.039 --> 00:23:25.079
Then there was support vector machines,
and then somebody else, and now then

336
00:23:25.119 --> 00:23:27.680
with deep learning, and now it's
AI. So sometimes we see people building

337
00:23:27.759 --> 00:23:33.119
workflows for even very simple things.
They're reaching out to some AI and we

338
00:23:33.319 --> 00:23:36.839
just say, hey, there's a
no denignment that does that computationally a lot

339
00:23:37.000 --> 00:23:38.759
less expensive. I don't use that. So I think, to me,

340
00:23:38.960 --> 00:23:42.079
it's it's a bit of a hype
right now. It's just a new kid

341
00:23:42.160 --> 00:23:45.640
in the block. Everybody wants to
play with it and use it. But

342
00:23:45.000 --> 00:23:49.440
the augmented really mixing it and matching
it right with traditional techniques, I think

343
00:23:49.480 --> 00:23:53.240
that's where the true value lies.
Yeah. Well, and so I'm just

344
00:23:53.359 --> 00:23:57.240
guessing here that one of the nice
things about your platform is that it is

345
00:23:57.400 --> 00:24:03.599
an end to end platform for building
models, designing models, training models,

346
00:24:03.640 --> 00:24:07.519
pulling the data in all these things, and it's adjacent to this business hub,

347
00:24:07.039 --> 00:24:14.279
So you have a marshaling area for
ideas and for testing algorithms and for

348
00:24:14.400 --> 00:24:18.200
testing models. Then you connect it
through the business hub and see what happens

349
00:24:18.279 --> 00:24:22.960
and see how it operates. And
it's important to have this one environment where

350
00:24:22.000 --> 00:24:26.400
that takes place because when you have
multiple tools, it just takes longer and

351
00:24:26.599 --> 00:24:32.799
it's disjointed and there are connections between
the tools and things change, So it's

352
00:24:32.880 --> 00:24:38.039
important to have that main marshaling area
to It's like a giant analytics sandbox.

353
00:24:38.160 --> 00:24:41.680
Is that about right? That's a
very nice sbscription. Absolutely. I tend

354
00:24:41.759 --> 00:24:49.000
to say that data scientists doesn't necessarily
need to know how the method does something,

355
00:24:49.160 --> 00:24:52.000
but it needs to know what the
method does. So if it's reaching

356
00:24:52.039 --> 00:24:55.920
out to a Python library or in
our library or sea libry underneath it,

357
00:24:55.960 --> 00:24:59.680
it's not that important, but you
still need to understand what the method actually

358
00:24:59.720 --> 00:25:03.839
does underneath it to be able to
interpret the results. It's a simple example.

359
00:25:03.839 --> 00:25:07.359
If you don't know what a regression
coefficient is, you won't be able

360
00:25:07.400 --> 00:25:10.960
to interpret it, but you don't
necessarily need to understand how it was derived

361
00:25:11.000 --> 00:25:14.079
from the data. Yeah, no, that's very interesting. Let me throw

362
00:25:14.119 --> 00:25:15.680
this concept at you and see what
you think about it. I wrote up

363
00:25:15.759 --> 00:25:19.720
an article just last week I guess
about this. I was flying to a

364
00:25:19.799 --> 00:25:23.920
conference in Denver just thinking about these
large language models and analytics and AI and

365
00:25:23.960 --> 00:25:27.079
all this stuff that have been covering
for a long long time, and I

366
00:25:27.160 --> 00:25:32.680
thought to myself about this concept I
call the executive cockpit. And the idea

367
00:25:32.759 --> 00:25:37.559
is that I think very forward looking
organizations are going to deploy a small language

368
00:25:37.640 --> 00:25:41.200
model that is aligned with their business, like if it's manufacturing or healthcare or

369
00:25:41.279 --> 00:25:47.599
whatever, in their data center,
so on prem, possibly in the cloud

370
00:25:47.599 --> 00:25:51.720
as well, but I have my
thoughts wrapped around this on prem small language

371
00:25:51.759 --> 00:25:55.920
model. Then you're going to train
it on your ERP, on your salesforce,

372
00:25:56.079 --> 00:25:59.839
on your CRM, on your customer
support for example, your tickets,

373
00:25:59.880 --> 00:26:03.640
like any of your core enterprise systems. You're going to train this model on

374
00:26:03.799 --> 00:26:07.680
your data, on your business business
data. Then what you'll do is set

375
00:26:07.759 --> 00:26:14.920
up copa topics coming from those systems
into a vector database adjacent to this interface

376
00:26:14.960 --> 00:26:18.880
for the small language model, and
that is where the executives will spend their

377
00:26:18.000 --> 00:26:22.359
day running their business. Because then
you could ask any question at all,

378
00:26:22.680 --> 00:26:26.039
how is our marketing working? In
APAC Who can we let go if we

379
00:26:26.160 --> 00:26:32.279
have to save some money? Where
are we weak in our organization right now?

380
00:26:32.359 --> 00:26:34.599
Just all kinds of different questions and
you'll get all these answers. And

381
00:26:34.680 --> 00:26:38.079
I actually mentioned to a CEO of
this one company because I was trying to

382
00:26:38.119 --> 00:26:41.400
get him to help me do sales
enablement for them, because I have this

383
00:26:41.480 --> 00:26:45.960
big audience I've been marketing to for
years. And one person turns out to

384
00:26:45.039 --> 00:26:51.480
be the next deputy Chief data officer
for the IRS. And I sent this

385
00:26:51.599 --> 00:26:53.519
email saying, Hey, this is
the lady I've known for a long long

386
00:26:53.640 --> 00:26:56.519
time. This is what I mean
by sales enablement. Do you guys have

387
00:26:56.640 --> 00:27:00.200
the IRS account? And he fire
back, he said, don't. I

388
00:27:00.200 --> 00:27:03.279
don't know for the other's accounts.
I thought to myself, well, you

389
00:27:03.400 --> 00:27:06.319
would know if you had the executive
cockpit, because you would just ask it,

390
00:27:06.480 --> 00:27:08.480
do we have the IRS account?
Who is the account rep? What's

391
00:27:08.519 --> 00:27:11.880
the latest of this account? Because
you're getting information from all these systems in

392
00:27:12.000 --> 00:27:15.319
your private environment. But what do
you think about this concept? Is that

393
00:27:15.920 --> 00:27:18.440
is that doable? Is that pie
in the sky or what do you think

394
00:27:18.480 --> 00:27:23.440
about all that? It's an interesting
idea, I thought about similar. I

395
00:27:23.480 --> 00:27:27.440
mean, at the end of the
day, you're personalizing a large language model

396
00:27:27.599 --> 00:27:33.799
around your own infrastructure in house data. I think the challenge there is that

397
00:27:33.039 --> 00:27:37.440
in order to get a really really
good model like that one that's really useful,

398
00:27:37.799 --> 00:27:40.599
you need to train it on a
lot more data than just your own.

399
00:27:41.440 --> 00:27:45.519
So in a sense, you need
to benefit from your competitor's data without

400
00:27:45.519 --> 00:27:48.039
actually seeing that, but kind of
learning the general structure and the general insights,

401
00:27:48.079 --> 00:27:52.279
and then you customize it on your
own, which in return kind of

402
00:27:52.400 --> 00:27:57.839
means that you should also be providing
your data to other organizations. It's almost

403
00:27:57.920 --> 00:28:03.599
like that's kind of pre competitive training
of these models so that they're useful for

404
00:28:03.720 --> 00:28:08.440
everybody. I think just training it
on your setup, you need some bigger

405
00:28:08.519 --> 00:28:12.559
context than that. Or maybe you're
a company and you have enough context anyway,

406
00:28:12.640 --> 00:28:15.079
But for every small company, I
don't think you have enough data to

407
00:28:15.160 --> 00:28:19.720
really get meaningful insights. That's very
interesting. That's a good that's a good

408
00:28:19.759 --> 00:28:23.480
point because I'm just I'm wondering to
myself and I'm gonna throw this one at

409
00:28:23.519 --> 00:28:26.839
you too. So one of my
AHA moments with these large language models is

410
00:28:26.880 --> 00:28:33.119
when I realized that when you train
them on a corpus of data. They're

411
00:28:33.200 --> 00:28:37.400
not actually persisting the data verbatim.
It's not like they're taking strings of text

412
00:28:37.480 --> 00:28:41.839
and storing it in a record somewhere, but rather, in the training process

413
00:28:41.440 --> 00:28:47.319
that data you use will adjust the
weights and biases and the parameters of the

414
00:28:47.480 --> 00:28:51.200
model. So in other words,
it's like, huh, well, that's

415
00:28:51.880 --> 00:28:56.960
that's very interesting that it can train
in that fashion and then reflect back to

416
00:28:56.039 --> 00:29:00.640
you such remarkably granular detail about things. And you know, what I've seen

417
00:29:00.720 --> 00:29:03.759
is that if there is a subject
area that has been published about widely,

418
00:29:04.400 --> 00:29:11.079
like how computer processors work, or
how an irrigation system works, anything that

419
00:29:11.200 --> 00:29:15.720
has a lot of content on the
web that these engines were trained on,

420
00:29:15.200 --> 00:29:18.119
it does very well. It knows
all that stuff. It's when you get

421
00:29:18.160 --> 00:29:22.160
to the fringe where there's not that
much published. And I guess that's kind

422
00:29:22.200 --> 00:29:26.160
of your point about having enough data
to train the models. If you don't

423
00:29:26.240 --> 00:29:29.640
have enough, you're not going to
get the contours right, and it's going

424
00:29:29.720 --> 00:29:33.039
to be skewed in one direction or
other. Is that about right? I

425
00:29:33.079 --> 00:29:34.319
think that's a very good summary.
The contrast is right. I mean,

426
00:29:34.599 --> 00:29:38.119
a colleague of mind want summarized,
since it essentially it's a consensus engine.

427
00:29:38.400 --> 00:29:44.400
It's getting the consensus around what a
computer programming is, learns that from the

428
00:29:44.480 --> 00:29:48.240
data and can repeat that. But
if it's just one isolated outcome, it's

429
00:29:48.559 --> 00:29:51.920
not going to be able to recall
that one. Interesting. Yes, So

430
00:29:52.319 --> 00:29:56.319
Craig schmid Huber I think his name
is. He's the guy who wrote the

431
00:29:56.359 --> 00:30:00.599
papers on the transformers, and he's
based I guess he's actually he Arabia these

432
00:30:00.680 --> 00:30:03.480
days, but I want to say
he's German of German origin. And I

433
00:30:03.640 --> 00:30:07.039
was amazed when I realized he wrote
those papers in like the nineteen nineties or

434
00:30:07.119 --> 00:30:11.720
something. And it's just just now
we have the compute to be able to

435
00:30:11.079 --> 00:30:14.920
can you explain that? Is that
what happened is that just the timing was

436
00:30:15.000 --> 00:30:18.359
right now to be able to understand
this and put it into play, because

437
00:30:18.440 --> 00:30:22.279
that was one of the big changes. And now it's able to see like

438
00:30:22.480 --> 00:30:25.960
you know, ten twelve tokens left
or right as opposed to just like two

439
00:30:26.079 --> 00:30:29.400
or three, and you also have
this, like you say, like a

440
00:30:29.519 --> 00:30:33.519
consensus right where so they are like
I call it almost like an ai Greek

441
00:30:33.640 --> 00:30:36.480
chorus where one is saying I think
it should be an A. I think

442
00:30:36.480 --> 00:30:37.119
it should be a B. I
think it should be a C. And

443
00:30:37.240 --> 00:30:41.920
then the Okay, I'm going to
pick this one. That's very interesting.

444
00:30:41.000 --> 00:30:45.039
It's a very interesting development. But
why do you think it took so long?

445
00:30:45.160 --> 00:30:48.359
Is it just because we now have
the compute to do that? I

446
00:30:48.400 --> 00:30:52.079
think it was a computer party issue
as well. And then some science tends

447
00:30:52.119 --> 00:30:55.559
to have a little bit more of
fun, needs a little bit of time

448
00:30:55.559 --> 00:30:59.160
before it truly has an impact,
but mostly waiting for complete power. I

449
00:30:59.200 --> 00:31:03.079
don't know that one way of looking
at what this consensus really does is I

450
00:31:03.079 --> 00:31:07.160
don't know if you watch these YouTube
videas about JGBD playing chess now and the

451
00:31:07.279 --> 00:31:11.559
interesting part is that at the beginning
it does extremely well and does very sensible

452
00:31:11.640 --> 00:31:15.960
things, and part of that is
these opening libraries are all over the place,

453
00:31:17.640 --> 00:31:22.200
so that's extremely well established consensus.
And then somewhere in the middle it

454
00:31:22.279 --> 00:31:26.559
starts inventing bizarre moves and suddenly new
figures pop up on the on the board

455
00:31:26.759 --> 00:31:30.119
out of nowhere, right, and
it has always meaningful explanations for that,

456
00:31:30.640 --> 00:31:34.680
And the problem there is that that
data is so sparse that there's no consensus

457
00:31:34.720 --> 00:31:38.279
to learn. So at the beginning
it sounds it almost looks like it understands

458
00:31:38.480 --> 00:31:42.559
chess rules. But the only reason
it does follow the chess rules is that

459
00:31:42.640 --> 00:31:47.920
they're so deeply ingrained in all of
the common material that you see that the

460
00:31:48.319 --> 00:31:51.839
kind of the likelihood of going outside
of the world book is too small.

461
00:31:52.279 --> 00:31:55.000
But somewhere in the middle of the
game it goes completely off the books.

462
00:31:55.440 --> 00:31:59.559
That's interesting, that's wild. So
one of my good friends in the business

463
00:32:00.039 --> 00:32:02.519
as a gentleman named Usama Fayad.
You may have come across from at some

464
00:32:02.559 --> 00:32:06.920
point. He was the first chief
data officer for Yahoo way back in the

465
00:32:07.000 --> 00:32:12.519
day, and now he runs the
Institute for Experiential AI over at Northeastern University

466
00:32:12.880 --> 00:32:15.279
here in the States. And I
had him in the show, and he's

467
00:32:15.400 --> 00:32:19.279
very funny, he's very candid.
He said, these large models, they're

468
00:32:19.359 --> 00:32:22.559
too big. They're not supposed to
work. We don't know why they work.

469
00:32:22.240 --> 00:32:24.960
What are you talking about, this
guy who runs this whole operation.

470
00:32:25.039 --> 00:32:28.839
He's joking, we don't even know
how they work. I mean, how's

471
00:32:28.880 --> 00:32:31.559
that for transparency, right, I
mean there's some truth, right. We

472
00:32:31.640 --> 00:32:35.640
don't really know how they come up
with these answers. Right, it's a

473
00:32:35.720 --> 00:32:38.079
wild mix of it. It's a
highly distributed model. We don't know why

474
00:32:38.200 --> 00:32:44.440
a particular answer comes. We can
come up with kind of proxies for an

475
00:32:44.440 --> 00:32:46.559
explanation by wiggling with the inputs and
trying to figure out what happens, and

476
00:32:46.680 --> 00:32:49.960
we can say, ah, this
probably had a lot of influence on the

477
00:32:50.079 --> 00:32:52.759
decision, but we don't know for
sure. That's so wild. I mean,

478
00:32:52.799 --> 00:32:55.119
that's just such a big deal that
you know, but we do.

479
00:32:55.480 --> 00:33:00.200
So now we have all this observability
in the data space, right, You've

480
00:33:00.240 --> 00:33:04.359
got Data Relic and Data Dog or
new Relative today Data. All these different

481
00:33:04.400 --> 00:33:08.119
companies are doing observability which I think
spun out of Kubernetes primarily. But it's

482
00:33:08.240 --> 00:33:12.880
very interesting and we need that kind
of observability in these large language models.

483
00:33:12.880 --> 00:33:15.200
I think. I think that's going
to be one of the keys to success.

484
00:33:15.640 --> 00:33:17.359
But folks don't touch that dot.
Will be right back. We're talking

485
00:33:17.400 --> 00:33:30.559
to Michael berthold from NIME on Inside
Analysis. Standby, Welcome back to Inside

486
00:33:30.599 --> 00:33:37.599
Analysis. Here's your host, Eric
Tabanac. All right, folks back here

487
00:33:37.640 --> 00:33:42.279
on Inside Analysis with Michael Bertholdt,
founder and CEO of nime K n I

488
00:33:42.559 --> 00:33:45.920
m e. Looked them up online
and Michael I was mentioning to you in

489
00:33:45.000 --> 00:33:51.640
the break that I'm wondering to myself
this whole business intelligence industry and there are

490
00:33:51.960 --> 00:33:55.440
hundreds of players these days, hundreds
of companies doing some form of analytics.

491
00:33:55.519 --> 00:34:00.079
Of course, NIME is an whole
analytics platform and open source platform end to

492
00:34:00.319 --> 00:34:05.240
end, but there are lots of
point tools, whether it's visualization or number

493
00:34:05.319 --> 00:34:07.199
crunching, OL app roll, app
all this kind of stuff, and I

494
00:34:07.360 --> 00:34:12.159
wonder is all of that in the
crosshairs of these foundational models? What do

495
00:34:12.199 --> 00:34:15.960
you think? That's a very very
interesting question, and we of course asked

496
00:34:16.039 --> 00:34:20.559
ourselves that as well. And I
think for some of the some of the

497
00:34:20.639 --> 00:34:23.800
tools that you mentioned, like generating
visualizations, that type of stuff, I

498
00:34:24.000 --> 00:34:29.559
do think they are pretty replaceable by
AI type models, because at the end

499
00:34:29.599 --> 00:34:34.639
of the day, you're doing something, you're generating a code that generates the

500
00:34:34.719 --> 00:34:37.760
visualization based on data, and you
judge the output of that code by just

501
00:34:37.840 --> 00:34:40.639
looking at it and saying this is
quite right. So I think that type

502
00:34:40.679 --> 00:34:45.639
of stuff will go away. And
we have in NIME actually built in what

503
00:34:45.880 --> 00:34:49.480
used to be an each chart scripting
editor that has now an AI element and

504
00:34:49.559 --> 00:34:52.519
you don't need to touch the code
anymore. So those types of wills I

505
00:34:52.639 --> 00:34:58.519
believe will go away. The eye
tools trying to really find surprising, interesting

506
00:34:58.679 --> 00:35:02.639
new insights in I think that type
of stuff is a lot harder to replace

507
00:35:04.039 --> 00:35:07.000
because fundamentally you're trying to find something
new. And like we discussed before the

508
00:35:07.079 --> 00:35:12.840
break, these GENEI models are consensus
engines, right, so they kind of

509
00:35:13.480 --> 00:35:17.519
try to gravitate towards something they've seen
more and more often before. Interesting.

510
00:35:17.679 --> 00:35:22.320
That's right, that's an excellent point. Really, that's that's exactly right.

511
00:35:22.679 --> 00:35:28.800
So it's good for understanding the well
trodden path basically, Like that's what it's

512
00:35:28.920 --> 00:35:30.199
very good at doing, is saying, Okay, there's a highway, it

513
00:35:30.280 --> 00:35:35.280
goes that direction, but I want
to go wandering around the forest, and

514
00:35:35.360 --> 00:35:38.320
it's not as good on the fringe
basically, so you will use it.

515
00:35:38.400 --> 00:35:42.599
But I mean, so I read
an article some guy on LinkedIn talking about

516
00:35:42.639 --> 00:35:47.119
how he connected I don't know by
ODBC or JDBC or something in his model

517
00:35:47.239 --> 00:35:51.320
with data sources, and he asked
it to queer the data source and it

518
00:35:51.400 --> 00:35:54.119
did. It reached into the database, pulled the information out and delivered it

519
00:35:54.159 --> 00:35:59.320
and you're like, Okay, that's
pretty interesting. And then when I think

520
00:35:59.360 --> 00:36:04.320
to myself, what's what could be
happening here? Is in the data warehousing

521
00:36:04.400 --> 00:36:07.199
space, for example, we move
so much data around. It's all the

522
00:36:07.320 --> 00:36:10.239
data that's from your core systems that
you've decided to put in, which is

523
00:36:10.239 --> 00:36:14.840
a tremendous amount. Very little of
that data ever gets used a lot of

524
00:36:14.920 --> 00:36:19.599
times it's the it's the summaries or
the aggregates or the roll ups that are

525
00:36:19.719 --> 00:36:22.760
used for various purposes, but a
lot of it just doesn't even get used

526
00:36:22.800 --> 00:36:25.880
at all. And I think that
what these large language models are going to

527
00:36:25.960 --> 00:36:30.679
do is kind of turn the entire
model inside out of how we viewed moving

528
00:36:30.880 --> 00:36:36.960
data and analyzing data and doing things
with data because they don't really care.

529
00:36:37.599 --> 00:36:40.119
They're just going to once they're trained
on a certain space. And again,

530
00:36:40.119 --> 00:36:43.840
if you train it on your data, or if you're in your vector database,

531
00:36:43.840 --> 00:36:45.760
you have a lot of embeddings of
your corporate data and you point your

532
00:36:45.840 --> 00:36:51.199
RAG model there, well, you
can get answers to things very quickly that

533
00:36:51.360 --> 00:36:54.960
before would have required running reports and
doing ETL and doing all this stuff.

534
00:36:55.239 --> 00:36:59.400
And I think that in many use
cases. These models are going to short

535
00:36:59.440 --> 00:37:01.360
circuit all all that stuff and you're
just not going to have to do as

536
00:37:01.400 --> 00:37:05.199
much that stuff anymore at all.
But what do you think about that?

537
00:37:06.920 --> 00:37:10.760
I think there's some truth to it, because fundamentally, what these models won't

538
00:37:10.920 --> 00:37:15.559
necessarily do is actually look at all
of the data, but they're going to

539
00:37:15.679 --> 00:37:19.760
apply a lot of common standard practices
to that. And standard sounds a little

540
00:37:19.760 --> 00:37:23.000
bit too limiting. I think there's
a huge wealth of standard practices that people

541
00:37:23.039 --> 00:37:27.679
do apply to the data, and
that's part of this consensus engine, and

542
00:37:27.840 --> 00:37:31.519
so that the AI models will try
out a lot of those things a lot

543
00:37:31.639 --> 00:37:36.320
faster than you ever would. So
absolutely, and there's a good chance that

544
00:37:36.400 --> 00:37:40.920
some of these insights that will be
generated are interesting to you. But then

545
00:37:42.079 --> 00:37:45.719
continuing the exploration and saying, I
mean, how do they always say the

546
00:37:45.559 --> 00:37:52.119
Eureka moment is usually preceded by oops, that's strange. I think you'll have

547
00:37:52.320 --> 00:37:55.880
these moments right, And AI doesn't
do that. AI doesn't say this is

548
00:37:55.960 --> 00:38:00.800
weird, I should diggle in a
little bit deeper because that's outside of the

549
00:38:00.880 --> 00:38:04.159
consensus. So it will continue doing
kind of like you said, will will

550
00:38:04.239 --> 00:38:13.440
continue the normal path and that's why
I believe the human intuition curiosity OOPS detection

551
00:38:13.559 --> 00:38:17.800
capability is going to be relevant for
a long time. I like this oops

552
00:38:17.880 --> 00:38:21.719
detection. That's good stuff. Well, there was a gentleman I had on

553
00:38:21.760 --> 00:38:23.800
the show years ago who did something. He said something a lot like that.

554
00:38:24.360 --> 00:38:28.400
He basically said, AI doesn't have
to be the ability to be like,

555
00:38:28.880 --> 00:38:31.440
hmm, that's kind of weird.
What's going on with that? Right?

556
00:38:31.519 --> 00:38:36.320
Because it's just processing information and doing
what it's been told to do,

557
00:38:36.400 --> 00:38:39.480
which is just reflect backwards based upon
a prompt and its training. It's a

558
00:38:39.559 --> 00:38:43.800
very simple thing. I mean,
it's very complex in terms of how it

559
00:38:43.960 --> 00:38:46.599
got there, but nonetheless it you
know, one thing that did annoy me,

560
00:38:46.679 --> 00:38:50.760
I will say is in the early
days when that New York Times reporter

561
00:38:51.480 --> 00:38:55.480
was getting deep with the with CHATGBT
and trying to like tease out of it

562
00:38:55.559 --> 00:38:59.880
whether it's sentient or something. I'm
like, dude, that is a misuse

563
00:39:00.119 --> 00:39:02.760
of the technology, Like that is
not whether you should be using this thing

564
00:39:02.880 --> 00:39:07.679
for to try to like what trick
it into revealing that it's really alive and

565
00:39:07.079 --> 00:39:10.079
you know, what are you even
talking about? And I think that's part

566
00:39:10.159 --> 00:39:15.760
of the downside these days is that. And I'm a media person myself,

567
00:39:15.840 --> 00:39:19.920
but a lot of times the media
will just sort of glom onto some narrative

568
00:39:19.960 --> 00:39:23.400
about something and it's very hard for
them to decouple from that and get down

569
00:39:23.440 --> 00:39:25.599
to brass tacks. And that's what
we do in the show. In fact,

570
00:39:25.639 --> 00:39:29.079
I used to say at the beginning
of every show, the show,

571
00:39:29.119 --> 00:39:31.800
it's all about getting down to the
brass tax of what actually happens in the

572
00:39:31.880 --> 00:39:35.719
data world and what you do with
this stuff. And I think it is

573
00:39:35.800 --> 00:39:39.239
important that people keep in their minds
the purpose of this technology, Why are

574
00:39:39.320 --> 00:39:43.199
you using it? Where is it
appropriate to use it, and where is

575
00:39:43.199 --> 00:39:45.199
it not appropriate to use it?
And that's just basic common sense, right,

576
00:39:46.159 --> 00:39:49.880
Yes, I totally agree. I
mean I go pretty much in line

577
00:39:49.920 --> 00:39:52.239
with also the European AII that they
just passed. I mean, if it's

578
00:39:52.719 --> 00:39:57.440
not mission critical, if it's not
safety critical, you can trust a system

579
00:39:57.559 --> 00:40:00.480
that is wrong in I don't know, point one percent of all cases.

580
00:40:00.639 --> 00:40:06.000
If it's controlling nuclear power plants,
better not be wrong in point one percent

581
00:40:06.039 --> 00:40:08.280
of all cases, right, that's
right. You got to watch out where

582
00:40:08.320 --> 00:40:12.559
it's so where do you see a
lot of use case of your clients.

583
00:40:12.599 --> 00:40:15.239
I mean, obviously some of your
clients are using large language models. Where

584
00:40:15.239 --> 00:40:21.840
are you seeing success stories in that
space right now? So there's a lot

585
00:40:21.880 --> 00:40:24.639
of success stories in other areas of
the business, as you probably probably know.

586
00:40:24.800 --> 00:40:29.960
Undoubtedly, no checking legal contracts,
doing marketing material, that type of

587
00:40:29.960 --> 00:40:34.199
stuff. There's a lot of value
in applying GENEI on the data analytic space.

588
00:40:34.840 --> 00:40:37.039
Honestly, we don't. We see
a lot of interest. There's a

589
00:40:37.079 --> 00:40:42.039
lot of people that say, oh
cool, I can build a customized chatboard

590
00:40:42.159 --> 00:40:45.920
using name that's not really our core
business. And then the real applications tend

591
00:40:46.000 --> 00:40:50.840
to be around text processing, which
is where jennai is really strong. And

592
00:40:51.239 --> 00:40:58.840
then instead of using outdated antique libraries
for sentiment analysis or text segmentation, you're

593
00:40:58.920 --> 00:41:00.960
just handing it over to an AI
model and say, hey, segment this,

594
00:41:01.199 --> 00:41:05.280
or extract the key components or create
a summary. But that type of

595
00:41:05.320 --> 00:41:09.679
stuff, it's amazing. So I
see. I'm also as image mining extensions.

596
00:41:09.800 --> 00:41:15.199
I think that's the next setup where
we can use image processing capabilities of

597
00:41:15.280 --> 00:41:19.760
Jennai for a lot of the number
crunching. I mean, we've all seen

598
00:41:19.800 --> 00:41:22.559
these cases where you can't add two
numbers doesn't really know what the prime numbers.

599
00:41:23.119 --> 00:41:28.760
This is the understanding of the concept
of a number. Right. So

600
00:41:28.880 --> 00:41:34.400
there. I think it's more as
a tool to help you build workflows,

601
00:41:34.559 --> 00:41:37.719
build dataations, but only as a
helper, right. So that's actually an

602
00:41:37.760 --> 00:41:42.880
excellent point I wanted to get into. I believe that we're just scratching the

603
00:41:42.960 --> 00:41:47.360
surface of using these models as a
component in a workflow. So you mentioned,

604
00:41:47.400 --> 00:41:52.639
for example summarization. That is hugely
powerful. I mean, you know

605
00:41:52.000 --> 00:41:58.639
you can enter especially for policies,
for complex policies for law, for example,

606
00:41:58.760 --> 00:42:04.280
for legal protocols and when to file
motions, what motions you can file,

607
00:42:04.360 --> 00:42:06.480
what you have to do according to
I mean, you used to have

608
00:42:06.559 --> 00:42:08.880
to pay lawyer's a lot of money
to tell you that stuff. Now if

609
00:42:08.920 --> 00:42:12.760
you just get access to the rules, will load them into a large language

610
00:42:12.760 --> 00:42:16.880
model and just start asking questions.
That is an incredibly powerful use case because

611
00:42:17.039 --> 00:42:21.079
it used to take a lot of
time to sort through the process of how

612
00:42:21.119 --> 00:42:23.239
to do something. Now you just
ask it, how do I fire someone?

613
00:42:23.559 --> 00:42:27.519
First step, send them a letter
saying they're not performing properly. Second

614
00:42:27.519 --> 00:42:29.960
step, you know, monitor their
behavior. Third step. With all this

615
00:42:30.000 --> 00:42:34.079
stuff, it's like there it is
like, Wow, talk about saving time.

616
00:42:34.199 --> 00:42:37.239
I mean, it saves time.
And here's another big soapbox issue.

617
00:42:37.280 --> 00:42:42.360
It improves morale because nobody wants to
spend their time scratching their head reading through

618
00:42:42.599 --> 00:42:46.000
just dreadful documentation. Nobody likes doing
that, nobody, So all that stuff

619
00:42:46.079 --> 00:42:50.280
is going to go away, right, what do you think? I totally

620
00:42:50.280 --> 00:42:52.679
agree. Yah. Your examples center
a lot around firing people, by the

621
00:42:52.719 --> 00:42:58.920
way, But I think I tend
to say, and people ask me if

622
00:42:59.239 --> 00:43:00.679
Jenny, I A, you going
to make data science lives easier? I

623
00:43:00.800 --> 00:43:04.880
say no, I don't think so. But it's going to make it nicer

624
00:43:04.960 --> 00:43:07.519
because it's going to remove all of
that boring stuff and we can now focus

625
00:43:07.599 --> 00:43:10.960
on the really interesting but more complex
stuff. So it's going to make it

626
00:43:12.039 --> 00:43:16.079
more interesting, more complex. That's
interesting. Yeah, wow, And I

627
00:43:16.119 --> 00:43:20.440
think you do want to document things, and you can have it document things

628
00:43:20.519 --> 00:43:22.440
for you too, Right, you
can just throw a whole bunch of stuff

629
00:43:22.639 --> 00:43:24.800
and document this, okay exactly.
I mean, we we now have a

630
00:43:24.840 --> 00:43:29.639
component on the hub that takes a
nine workflow and explains what the nine workflow

631
00:43:29.679 --> 00:43:31.239
does, and we do that by
just shipping it off to Jennai, it's

632
00:43:31.280 --> 00:43:35.400
perfect for that. Wow, that's
amazing. All right, folks, Well

633
00:43:35.519 --> 00:43:42.599
podcast p A segment coming up next. We're listening to Inside Analysis. All

634
00:43:42.679 --> 00:43:45.679
right, folks back here on Inside
Analysis talking to Michael Bertholdt is the founder

635
00:43:45.719 --> 00:43:51.639
and CEO of nine k N I
m and Michael, I know what it's

636
00:43:51.760 --> 00:43:53.440
like in a software company. There's
always a roadmap. You're always working on

637
00:43:53.559 --> 00:43:59.199
something, and we talked about a
couple of key things. Governance. There's

638
00:43:59.320 --> 00:44:01.800
model governance, there's data governance,
there's it governance. What are you working

639
00:44:01.840 --> 00:44:07.239
on in the governance space? Thanks
for asking that, Like, like you

640
00:44:07.320 --> 00:44:10.639
know, all the road maps are
changing all the time. But what we're

641
00:44:10.679 --> 00:44:15.360
currently working on. We had actually
this model government's topic on the work right,

642
00:44:15.440 --> 00:44:16.920
brend I've been working on that for
a couple of years now. So

643
00:44:17.079 --> 00:44:22.159
the idea of being able to monitor
what models are doing automatically we train them.

644
00:44:22.199 --> 00:44:27.000
We talked about that and I think
the first episode, but what we

645
00:44:27.119 --> 00:44:31.360
added now is the ability to also
govern the AI usage of people that are

646
00:44:31.400 --> 00:44:36.119
creating nine work clows. So first
of all, then somebody is creating nine

647
00:44:36.159 --> 00:44:39.360
workplos using the NIME analytics platform and
uses this built in AI. We call

648
00:44:39.400 --> 00:44:44.559
it KAI for NIME AI. We
need to make sure that gets channeled to

649
00:44:44.679 --> 00:44:47.719
an IT approved AI. Right,
maybe that's just for expense purposes. You

650
00:44:47.760 --> 00:44:50.920
know, I'm going to have too
much consumption in the cloud. Want you

651
00:44:51.039 --> 00:44:53.920
make that in house or it's really
a data privacy issue. But the more

652
00:44:54.079 --> 00:44:59.119
worrisome part for people is that you
I mean, one of the springs of

653
00:44:59.199 --> 00:45:02.199
the NIE Analytics platform, the workflow
concepts, is that everybody can use any

654
00:45:02.280 --> 00:45:06.599
technology they want. Right, they
can reach out to experimental libraries, they

655
00:45:06.639 --> 00:45:08.960
can reach out to our stuff,
to Python stuff, to whatever. But

656
00:45:09.159 --> 00:45:15.599
by now they can of course also
connect to various different AI providers, and

657
00:45:15.719 --> 00:45:20.239
we need a way for them for
Central Light Tea Governance to be able to

658
00:45:20.719 --> 00:45:27.239
make sure that the nine workflow users
inside the organization can only use approved AIS.

659
00:45:27.639 --> 00:45:30.360
So maybe the maybe marketing can use
an AI in the cloud, but

660
00:45:30.480 --> 00:45:35.519
maybe legal shouldn't or HR shouldn't,
right, And that's something we have built

661
00:45:35.559 --> 00:45:39.039
in into the name hub now that
we can limit the types of AIS,

662
00:45:39.199 --> 00:45:45.239
you can their users can reach out
to from the nine Ndredrix platform and they

663
00:45:45.320 --> 00:45:49.719
get to choose from one of the
approved AIS that it central Light he set

664
00:45:49.800 --> 00:45:52.639
up and said, okay, here's
an AI that's consumption light, that's for

665
00:45:52.719 --> 00:45:55.400
the easy tasks. Here's the one
for whatever the tech team. Here's the

666
00:45:55.480 --> 00:46:00.639
one that's for compliant data. And
we also allow they're set up on the

667
00:46:00.719 --> 00:46:05.880
hubside of safeguarding workflows so that you
can before the data gets sent out to

668
00:46:05.960 --> 00:46:09.679
say a cloud AI provider, it
gets screened for private information or maybe the

669
00:46:09.760 --> 00:46:15.199
data automatically it gets anonymized before it
gets sent out. Yeah, that's very

670
00:46:15.199 --> 00:46:21.079
important stuff. And the use are
you also able to do some finops on

671
00:46:21.199 --> 00:46:23.760
that, In other words, see
how much it is costing to leverage this

672
00:46:23.840 --> 00:46:28.440
AI engine versus that AI engine and
do some cost optimization. Is that something

673
00:46:28.480 --> 00:46:31.679
you can do. We can do
that as part of a nine workflow and

674
00:46:31.679 --> 00:46:35.199
you could build that. You could
build that in there as well. But

675
00:46:35.280 --> 00:46:38.719
we are currently offering to our customers
the abilities to monitor our consumption so they

676
00:46:38.800 --> 00:46:43.079
have a bit of an eye on
that. But it's not automatically re routing

677
00:46:43.159 --> 00:46:46.719
to different AIS. But that's just
an added functionality under the wood m hmm.

678
00:46:46.960 --> 00:46:50.639
Yeah, it's all in the workflow
is basically, and that's where we're

679
00:46:50.719 --> 00:46:53.480
going. In the last segment where
I think that we're just the beginning of

680
00:46:53.639 --> 00:46:59.920
leveraging these technologies because what they're really
very good at is pattern recognition, right

681
00:47:00.079 --> 00:47:04.440
even just the vectorization the embedding is
basically how it stores it as a point

682
00:47:04.480 --> 00:47:08.840
in array basically, and just understanding
how it can map those two things.

683
00:47:08.920 --> 00:47:13.199
It's not just word not just text
generation. I think we're going to get

684
00:47:13.239 --> 00:47:17.400
some really interesting things in terms of
pattern recognition and then recommendations. I mean,

685
00:47:17.440 --> 00:47:22.039
I think that these little AI agents, these assistants are going to be

686
00:47:22.159 --> 00:47:25.280
extremely helpful in all facets of business. You know, to be able to

687
00:47:25.960 --> 00:47:30.159
very quickly give you a customer profile
when you're on the phone with someone,

688
00:47:30.199 --> 00:47:34.440
or to be able to give you
summarization of text on demand. I mean,

689
00:47:34.519 --> 00:47:37.079
really, I think the hardest challenge
is going to be changing mindsets and

690
00:47:37.239 --> 00:47:40.679
changing day to day behaviors and workflows. What do you think final thoughts?

691
00:47:42.920 --> 00:47:45.639
I totally agree with that. But
we see that inside NIM as well.

692
00:47:45.719 --> 00:47:47.280
I mean, the developers were the
first ones that really said we want to

693
00:47:47.360 --> 00:47:50.920
use this, you want to use
this, but then getting the rest of

694
00:47:51.000 --> 00:47:53.159
the organization to as seriously think about
It's like what can HR do? But

695
00:47:53.239 --> 00:47:57.400
can legal really do with that.
It's a huge time saver, and it's

696
00:47:57.559 --> 00:48:01.719
largely untipped. I totally with you
as a check and they're interesting times a

697
00:48:01.800 --> 00:48:05.960
hit yes, to say the least. Well, what a fantastic conversation,

698
00:48:06.000 --> 00:48:08.480
folks, look them up online.
Michael Berthold B E R T H O

699
00:48:08.719 --> 00:48:12.840
L D from nine K N I
M E. We'll talk to you next

700
00:48:12.880 --> 00:48:19.679
time. Folks, you've been listening
to inside ANALLISONA. It's my fatous,

701
00:48:19.840 --> 00:48:25.840
cool logic radio legend you love and
the best talk progressive talk in southern California.

702
00:48:27.599 --> 00:48:35.199
We listen to you all the time. More than eighty million Americans depend

703
00:48:35.280 --> 00:48:38.559
on AM radio each month for news, weather and emergency information. A new

704
00:48:38.639 --> 00:48:43.920
bill in Congress would make sure AM
radio remains in cars because when sell and

705
00:48:44.000 --> 00:48:47.599
internet services are down, this free
service could be your only lifeline. Text

706
00:48:47.679 --> 00:48:52.840
AM to five two eight eighty six
and tell Congress to support the AM radio

707
00:48:52.960 --> 00:48:55.440
for every vehicle. ACT message in
better rates may applay. You may receive

708
00:48:55.519 --> 00:48:59.159
up to four messages a month,
and you may text stop to stop.

709
00:48:59.400 --> 00:49:05.800
This message furnished by the National Association
of Broadcasters. Tune into the Farran Dozier

710
00:49:05.960 --> 00:49:09.400
Show usually marks place in time,
The soundtrack to Life. Sunday Nights at

711
00:49:09.480 --> 00:49:15.920
eight pm are KCAA Radio playing the
hottest hits and the coolest conversations Sunday Nights

712
00:49:16.000 --> 00:49:21.480
at APM on The Ferran Dozier Show
within the array of music, talk,

713
00:49:21.920 --> 00:49:28.039
sports, community outreach and veteran resources, the hits from the sixties, seventies,

714
00:49:28.440 --> 00:49:34.280
eighties, nineties and today's hits.
The Faran Dozier Show on KCAA Radio

715
00:49:35.039 --> 00:49:40.400
on all available streaming platforms and on
a six point FiveM and ten fifty am

716
00:49:40.800 --> 00:50:00.599
The Faran Dozier Show on KCAA Radio. Redlands Ranch Marketing is a unique,

717
00:50:00.639 --> 00:50:06.280
full surface international grocery store that specializes
in authentic food items from Mexico, India,

718
00:50:06.480 --> 00:50:10.159
and from many Mediterranean and Asian countries, including popular items from the US.

719
00:50:10.360 --> 00:50:15.400
They offer fresh baked items from their
in house fagery, housemade tortillas from

720
00:50:15.440 --> 00:50:20.039
their tortilla area, a delicious array
of prepared Mexican foods, a terrific fresh

721
00:50:20.119 --> 00:50:22.800
food and juice bar, and a
large selection of meats, seafoods and deli

722
00:50:22.920 --> 00:50:28.320
sandwiches, salads and halal meats.
Their produce department is stocked full with fresh

723
00:50:28.400 --> 00:50:32.400
local and hard to find international fruits
and vegetables that you cannot find anywhere else.

724
00:50:32.639 --> 00:50:37.360
Don't forget to step into the massive
beer Cave and experience the largest selection

725
00:50:37.519 --> 00:50:42.920
of domestic, artisan and imported beers
in the IE. They can also cater

726
00:50:43.039 --> 00:50:46.079
your next event with one of the
delicious takeout catering trades of food. Visit

727
00:50:46.159 --> 00:50:51.639
them at Redlands ranch Market dot com. That's Redlands ranch Market dot com.

728
00:50:51.840 --> 00:50:59.360
Redlands Ranch Market a unique and fun
shopping destination. What if we took monopoly

729
00:50:59.599 --> 00:51:04.039
out of our our daily diets?
Busting the power of abusive and arrogant food

730
00:51:04.119 --> 00:51:07.960
monopolis would be of such immediate benefit
to the bottom lines of farmers, consumers,

731
00:51:08.000 --> 00:51:13.639
and workers that even middle of the
road congressional Democrats and a few Republicans

732
00:51:13.920 --> 00:51:19.440
are turning into ROOSEVELTI and trustbusters.
New Jersey Senator Cory Booker, for example,

733
00:51:19.840 --> 00:51:23.719
usually a reliable defender of corporate interests, sees the connection between inner city

734
00:51:23.800 --> 00:51:30.119
food deprivation and the consolidation of power
by industrial farm and food profiteers, which

735
00:51:30.239 --> 00:51:36.519
he says are contrary to our very
idea of farming in our country. Working

736
00:51:36.639 --> 00:51:42.079
with progressive grassroots groups like Food and
Water Action and Family Farm Action. Booker

737
00:51:42.199 --> 00:51:47.320
is sponsoring the Farm System Reform Act, a comprehensive proposal to overhaul major parts

738
00:51:47.360 --> 00:51:52.480
of the broken food structure included in
his bill, or strong overdue provisions to

739
00:51:52.559 --> 00:51:59.920
phase out the monstrous system of cafoes, confined animal feeding operations that are tortuous

740
00:52:00.280 --> 00:52:04.280
holes for thousands of chickens, hogs, and other animals caged in each of

741
00:52:04.360 --> 00:52:07.480
these huge factories, which are also
polluters of water, air, and rural

742
00:52:07.559 --> 00:52:14.559
communities. Rather than conventional liberal programs
to treat the symptoms of monopoly, such

743
00:52:14.639 --> 00:52:20.239
progressive populist approaches begin to dismantle monopoly, and they represent our best chance of

744
00:52:20.320 --> 00:52:24.000
actually making life fair for the majority
of people. The good news is that

745
00:52:24.159 --> 00:52:30.360
much of the power to do this
already exists. As investigative reporter Amy Swan

746
00:52:30.519 --> 00:52:35.599
writes in the January Washington Monthly,
we don't have to wait on recalcitrant Republicans,

747
00:52:35.679 --> 00:52:39.000
and we need Dems in Congress to
make progress. A tool shed of

748
00:52:39.119 --> 00:52:44.440
laws that were put in place during
the past one hundred years to counter monopoly

749
00:52:44.559 --> 00:52:49.079
power are still on the books.
They're stored in the agg Department, sec

750
00:52:49.679 --> 00:52:53.519
Justice Department, and many other drawers
of public power, mister Jim Hiar saying

751
00:52:53.800 --> 00:52:59.920
let's put them to work. Need
insurance help The insurance all with Carl Susman

752
00:53:00.119 --> 00:53:04.320
is your answer from policy queries to
coverage concerns. Carl and acclaimed expert in

753
00:53:04.400 --> 00:53:07.320
media, guest offer solutions. Join
us for a show that simplifies insurance,

754
00:53:07.719 --> 00:53:15.639
making it understandable for everyone. To
Hebot Club's original pure powdy rcosuper Ta comes

755
00:53:15.679 --> 00:53:17.840
from the only tree in the world
that fungus does not grow on. As

756
00:53:17.920 --> 00:53:22.679
a result, it naturally has anti
fungal, anti infection, anti viral,

757
00:53:22.719 --> 00:53:27.679
antibacterial, anti inflammation, and anti
parasite properties. So the TA is great

758
00:53:27.719 --> 00:53:30.280
for healthy people because it helps build
the immune system, and it can truly

759
00:53:30.360 --> 00:53:35.800
be miraculous for someone fighting a potentially
life threatening disease due to an infection,

760
00:53:36.039 --> 00:53:39.719
diabetes, or cancer. The T
is also organic and naturally caffeine free.

761
00:53:39.920 --> 00:53:44.599
A one pound package of T is
forty nine ninety five which includes shipping.

762
00:53:44.880 --> 00:53:47.840
To order, please visit to ebot
club dot com. To hebo is spelled

763
00:53:47.920 --> 00:53:52.960
T like tom, a h ee
B like boy oh. Then continue with

764
00:53:53.000 --> 00:53:58.000
the word T and then the word
club. The complete website is to Hebot

765
00:53:58.159 --> 00:54:01.400
club dot Com or call us at
eight one eight sixty one zero eight zero

766
00:54:01.559 --> 00:54:07.000
eight eight Monday through Saturday, nine
am to five pm California time. That's

767
00:54:07.079 --> 00:54:10.559
eight one eight sixty one zero eight
zero eight eight t ebot club dot Com

768
00:54:12.360 --> 00:54:20.079
with sixty years of fascinating facts.
This is the man from yesterday and back

769
00:54:20.119 --> 00:54:23.039
in time. We go to this
time in nineteen seventy two, Cynthia Lennon

770
00:54:23.320 --> 00:54:29.719
claims that John Lennon hasn't exercised his
visitation rights. They have a nine year

771
00:54:29.760 --> 00:54:34.599
old son named Julian, and according
to Cynthia, John Lennon hasn't seen his

772
00:54:34.719 --> 00:54:38.360
son Julian in about eight months.
This is com Le's go to get you

773
00:54:40.719 --> 00:54:51.920
go to nug you rot ahead,
You're better get yourself together and from this

774
00:54:52.079 --> 00:54:55.920
time in two thousand, Dennis Miller
is chosen as one of two new commentators

775
00:54:57.119 --> 00:55:02.159
to work alongside Al Michaels on ABCTV
Monday Night Football. Not only has Lindell

776
00:55:02.280 --> 00:55:07.400
never kicked on a Sunday, He's
never kicked a football in his life.

777
00:55:09.000 --> 00:55:14.360
Never. And from about this time
in nineteen fifty seven, Robert Young says

778
00:55:14.440 --> 00:55:17.559
part of the success of his father
knows best is that the public likes a

779
00:55:17.639 --> 00:55:30.199
dad who isn't of Afoon and Jane
White with Eleanor Donahue, Billy Gray and

780
00:55:30.360 --> 00:55:38.440
Lauren Chapin in Modern Knows Best with
more at Man from Yesterday dot com.

781
00:55:43.239 --> 00:55:47.039
Do you like to safely leverage bank
money to earn double digit returns income tax

782
00:55:47.159 --> 00:55:52.239
free, with guarantees and no downside
market risk? How can you do this?

783
00:55:52.559 --> 00:55:55.039
This is Farence, host of the
Your Personal Bank Show. One you

784
00:55:55.199 --> 00:56:00.880
fund a high cash value policy one
time to earn dividends and interest. Two

785
00:56:00.280 --> 00:56:05.360
establish a bank line of credit using
the cash in your policy as collateral.

786
00:56:05.519 --> 00:56:08.400
When you earn more in dividends from
your policy than the interest the bank charges,

787
00:56:08.639 --> 00:56:13.079
you keep the difference, and the
difference is average two to five percent

788
00:56:13.199 --> 00:56:16.719
annually in your favor for the past
forty plus years. Three the bank funds

789
00:56:16.800 --> 00:56:22.119
contributions years two to twenty plus.
Each year the bank adds funds, your

790
00:56:22.199 --> 00:56:25.039
rate of return increases. Your average
rate of return can grow too strong double

791
00:56:25.079 --> 00:56:30.840
digits annually within a few years.
Contact niyat your Personal bank dot com Your

792
00:56:30.920 --> 00:56:35.239
Personal bank dot com or eight sixty
six two six eight four four two two

793
00:56:35.440 --> 00:56:38.960
eight sixty six two six eight four
four two two for more info or tune

794
00:56:39.039 --> 00:56:44.559
in to the Your Personal Bank Show. Your Personal Bank Show airs Tuesdays at

795
00:56:44.559 --> 00:56:49.039
four pm right here on CASEAA ten
fifty AM and one oh six point five

796
00:56:49.159 --> 00:56:53.519
FM, the station that leaves no
listeners behind. Palm Springs Dispensary reminds everyone

797
00:56:53.639 --> 00:56:59.199
during these challenging times they understand the
importance of mental health. Offering daily deals

798
00:56:59.239 --> 00:57:02.639
on everything from time quality flowers to
edibles and discounts on CBD products. Palm

799
00:57:02.639 --> 00:57:06.920
Springs Dispensary's goal is to help you
win the battle within. As one of

800
00:57:06.920 --> 00:57:10.159
the area's most beautiful wellness shops.
Located on Garnet Avenue, visit the Palm

801
00:57:10.199 --> 00:57:16.039
Springs Dispensary dot com. That's the
Palm Springs Dispensary dot com. If you're

802
00:57:16.079 --> 00:57:20.360
looking for a fuller, part time
sales position and you have radio, TV

803
00:57:20.519 --> 00:57:23.599
or print media experience, k c
AA has a great opportunity waiting for you

804
00:57:23.719 --> 00:57:27.840
that pays the highest commissions in the
market. If you're interested in a sales

805
00:57:27.920 --> 00:57:35.119
position with us, email CEO at
KCAA Radio dot com. KCAA listing the

806
00:57:35.199 --> 00:57:38.360
k c AA Lowolinda at one O
six point five FM K two ninety three

807
00:57:38.440 --> 00:57:45.440
c F Burno Valley, NBC News
Radio. I'm Chris Caragio. Former President

808
00:57:45.519 --> 00:57:49.840
Trump says he thinks there will be
a breaking point for his supporters if he

809
00:57:49.960 --> 00:57:52.480
sentenced to prison time in his hush
money case. In an interview with Fox

810
00:57:52.559 --> 00:57:55.400
and Friends Weekend, Trump said he
doesn't know if the public would stand for

811
00:57:55.519 --> 00:57:59.400
it if he were put behind bars. I think I think it would be

812
00:57:59.480 --> 00:58:01.039
tough for the p to take.
You know, at a certain point,

813
00:58:01.039 --> 00:58:05.960
there's a breaking point. The former
president's comments come just days after a Manhattan

814
00:58:06.000 --> 00:58:09.280
jury convicted him on all thirty four
counts of falsifying business documents. Israel's war

815
00:58:09.400 --> 00:58:14.079
cabinet is meeting to discuss President Biden's
plan to wind down the war in Gaza.

816
00:58:14.239 --> 00:58:17.159
The session comes after an aide for
Israeli Prime Minister Benjamin net Yahoo said

817
00:58:17.199 --> 00:58:22.920
his government is ready to accept the
framework for the proposal, despite reservations about

818
00:58:22.920 --> 00:58:27.320
the deal. The chief foreign policy
advisor to Netan Yahoo told Britain Sunday Times

819
00:58:27.440 --> 00:58:30.920
Israel dearly wants to see the release
of hostages held by Hamas. The official

820
00:58:30.960 --> 00:58:35.400
said, while it's not a good
deal Israel, we'll accept it. Cruis

821
00:58:35.440 --> 00:58:38.840
are making progress and restoring drinking water
to Atlanta area residents following a major at

822
00:58:38.880 --> 00:58:43.639
water main break. That's the word
from city officials as Atlanta remains under a

823
00:58:43.679 --> 00:58:47.000
state of emergency after a massive water
main break left parts of downtown without water

824
00:58:47.119 --> 00:58:52.119
this weekend. It also led to
boil water advisories and prompted businesses and parts

825
00:58:52.159 --> 00:58:58.039
of the city's downtown entertainment district to
close or cancel events. At least one

826
00:58:58.079 --> 00:59:00.360
person is dead and over two dozen
injured after a shooting at a birthday party

827
00:59:00.400 --> 00:59:06.039
in Ohio. The Akron Police Department
says officers responded to a report of shots

828
00:59:06.079 --> 00:59:08.760
fired in a neighborhood in East Akron
early this morning. Authority say up to

829
00:59:08.800 --> 00:59:13.559
thirty people were injured. One man
was killed. An investigation is ongoing.

830
00:59:13.960 --> 00:59:19.159
A raging brush fire prompted evacuations in
central California. State officials say the Coral

831
00:59:19.280 --> 00:59:23.039
fire started yesterday afternoon east of San
Francisco and is blown up to over twelve

832
00:59:23.119 --> 00:59:28.039
thousand acres. As of a late
morning, the blaze was only about fifteen

833
00:59:28.079 --> 00:59:31.760
percent contained. Officials in San Joaquin
County have set up an evacuation center in

834
00:59:31.840 --> 00:59:36.360
the town of Tracy and were urging
residents who lived near the fire to evacuate.

835
00:59:36.519 --> 00:59:43.920
I'm Chris Caragio, NBC News Radio, NBC News on KCAA Lowland sponsored

836
00:59:43.960 --> 00:59:49.360
by Teamsters Local nineteen thirty two,
Protecting the Future of Working Families Teamsters nineteen

837
00:59:49.440 --> 00:59:59.320
thirty two dot org. You're listening
to an ongore presentation of this program kcaas

838
00:59:59.639 --> 01:00:02.440
thank you to the near for this
edition of Justice Watch with Attorney Zulu Ali.

839
01:00:02.639 --> 01:00:07.239
I am Attorney Zulu Ali with a
Justice Watch crew, Rosa Nunnez,

840
01:00:07.400 --> 01:00:08.400
Michael malau Klaw