WEBVTT

1
00:00:00.360 --> 00:00:03.439
<v Speaker 1>So we've made progress on pretty much every front, and

2
00:00:03.520 --> 00:00:07.200
<v Speaker 1>I think it's worth it's worth hearing about that progress

3
00:00:07.200 --> 00:00:11.640
<v Speaker 1>from the team, and they will also ask answer questions,

4
00:00:11.720 --> 00:00:14.080
<v Speaker 1>and I'll look at my x feed as well see

5
00:00:14.080 --> 00:00:16.199
<v Speaker 1>if there's any zingers in there that I can answer.

6
00:00:17.039 --> 00:00:19.800
<v Speaker 1>But the I guess the main message we wanted to

7
00:00:19.800 --> 00:00:24.239
<v Speaker 1>convey is that the advertising technology for X only Twitter

8
00:00:24.719 --> 00:00:29.640
<v Speaker 1>has improved dramatically so that the targeting is actually appropriate.

9
00:00:29.679 --> 00:00:33.880
<v Speaker 1>So previously there really wasn't any targeting, so people basically

10
00:00:33.880 --> 00:00:37.000
<v Speaker 1>see spam ads meaning and a spam ad is an

11
00:00:37.000 --> 00:00:40.759
<v Speaker 1>ad that is that doesn't meet the user's needs, whereas

12
00:00:40.799 --> 00:00:43.679
<v Speaker 1>an AD that is that meets the user's needs if

13
00:00:43.679 --> 00:00:47.280
<v Speaker 1>it's a product or service that they want, especially when

14
00:00:47.320 --> 00:00:51.200
<v Speaker 1>they want it, that is actually content. So our goal

15
00:00:51.439 --> 00:00:54.359
<v Speaker 1>is to move as as quickly as possible to having

16
00:00:54.439 --> 00:00:59.479
<v Speaker 1>ads that for products and services that directly meet the

17
00:00:59.520 --> 00:01:02.679
<v Speaker 1>needs of you users. And we're looking for actually very

18
00:01:02.679 --> 00:01:09.400
<v Speaker 1>specific product ads. So we certainly welcome we welcome brand advertising,

19
00:01:09.439 --> 00:01:13.560
<v Speaker 1>we welcome sort of general message advertising. But the thing

20
00:01:13.599 --> 00:01:17.519
<v Speaker 1>we're actually asking advertisers to do is to is to

21
00:01:17.560 --> 00:01:21.560
<v Speaker 1>now consider doing ads for very specific products and services

22
00:01:22.000 --> 00:01:26.120
<v Speaker 1>so we can actually connect users with products and services.

23
00:01:25.840 --> 00:01:27.560
<v Speaker 2>That I would make their lives better.

24
00:01:28.280 --> 00:01:31.519
<v Speaker 1>And that's we finally have the technology to do that

25
00:01:32.120 --> 00:01:33.519
<v Speaker 1>by using AI matching.

26
00:01:34.159 --> 00:01:34.680
<v Speaker 2>Essentially, we.

27
00:01:36.200 --> 00:01:38.799
<v Speaker 1>Look at what users do publicly, you know, what, what

28
00:01:38.799 --> 00:01:40.719
<v Speaker 1>do they interact with, what do they like, what they've

29
00:01:41.560 --> 00:01:46.599
<v Speaker 1>what posts they view, and and then that creates essentially

30
00:01:46.680 --> 00:01:49.959
<v Speaker 1>the set of interests of the user. And then the

31
00:01:50.000 --> 00:01:53.079
<v Speaker 1>same thing with the ad. We see what what users

32
00:01:53.359 --> 00:01:56.680
<v Speaker 1>pay attention to what ads and then and that essentially

33
00:01:56.760 --> 00:02:00.200
<v Speaker 1>creates a profile for that ad, and we correlate the

34
00:02:00.280 --> 00:02:04.079
<v Speaker 1>user interests with that particular product of service, and that

35
00:02:04.519 --> 00:02:09.719
<v Speaker 1>results in radically better matching. And but this year, a

36
00:02:09.759 --> 00:02:12.719
<v Speaker 1>number of times I've talked to people and I've said

37
00:02:12.759 --> 00:02:15.000
<v Speaker 1>how many times for how many years have you used

38
00:02:15.639 --> 00:02:16.240
<v Speaker 1>x RA Twitter?

39
00:02:16.919 --> 00:02:18.879
<v Speaker 2>And they might say as long as ten years.

40
00:02:19.159 --> 00:02:22.319
<v Speaker 1>And I'll say how much have you bought on x

41
00:02:22.639 --> 00:02:25.000
<v Speaker 1>in even going back to when it was Twitter, and

42
00:02:25.080 --> 00:02:28.599
<v Speaker 1>they'd say almost nothing, if not nothing. And we want

43
00:02:28.639 --> 00:02:32.360
<v Speaker 1>to change that completely to where you can even go

44
00:02:32.439 --> 00:02:36.680
<v Speaker 1>on x and you look forward to the ads where

45
00:02:36.680 --> 00:02:39.960
<v Speaker 1>you're actually interested to see what ads show up, because

46
00:02:40.280 --> 00:02:42.719
<v Speaker 1>they're going to be for products and services.

47
00:02:42.159 --> 00:02:42.800
<v Speaker 2>That you want.

48
00:02:43.599 --> 00:02:48.479
<v Speaker 1>And and that's that's the sort of nirvana goal that

49
00:02:48.560 --> 00:02:52.080
<v Speaker 1>we have here, and we're using grock AI to help

50
00:02:52.120 --> 00:02:52.879
<v Speaker 1>achieve that goal.

51
00:02:53.479 --> 00:02:58.840
<v Speaker 2>That's that's really fundamental. And then we've also been asked.

52
00:02:58.599 --> 00:03:01.479
<v Speaker 1>About as related to grock, if somebody has a query

53
00:03:01.479 --> 00:03:04.719
<v Speaker 1>on groc, should we perhaps have some advertising there. We

54
00:03:04.879 --> 00:03:07.879
<v Speaker 1>we haven't done that yet, but we are working on that.

55
00:03:08.719 --> 00:03:12.439
<v Speaker 1>But yeah, Number number one request of advertisers is very simple,

56
00:03:12.479 --> 00:03:16.719
<v Speaker 1>which is to put product and service, very very specific products,

57
00:03:17.120 --> 00:03:20.319
<v Speaker 1>very specific services on the system so that GROC can

58
00:03:20.360 --> 00:03:24.280
<v Speaker 1>match the your products and services to you know, the

59
00:03:24.319 --> 00:03:27.400
<v Speaker 1>six hundred million monthly users that we have on the system,

60
00:03:27.960 --> 00:03:30.400
<v Speaker 1>and when you when you consider total number of users

61
00:03:30.800 --> 00:03:33.039
<v Speaker 1>on on the X system, it's closer to a billion.

62
00:03:33.560 --> 00:03:36.479
<v Speaker 1>Some only will come in every you know, three months

63
00:03:36.560 --> 00:03:39.479
<v Speaker 1>or something like that, but total live people with accounts

64
00:03:39.680 --> 00:03:42.599
<v Speaker 1>is about a billion on the X platform. And it's

65
00:03:42.639 --> 00:03:46.240
<v Speaker 1>also very much worth emphasizing that the users on the

66
00:03:46.400 --> 00:03:52.560
<v Speaker 1>X platform have the highest buying power of any audience anywhere, so,

67
00:03:53.879 --> 00:03:55.520
<v Speaker 1>you know, so certainly have a lot of respect for

68
00:03:55.560 --> 00:03:59.960
<v Speaker 1>the advertising platforms on you know, TikTok or even snap

69
00:04:00.080 --> 00:04:03.599
<v Speaker 1>Chat and others, but very often the user base is

70
00:04:03.680 --> 00:04:06.199
<v Speaker 1>very young and does not have the money to buy things.

71
00:04:06.639 --> 00:04:11.520
<v Speaker 1>Our our user base consists of the smartest and wealthiest

72
00:04:11.520 --> 00:04:15.080
<v Speaker 1>people in the world, so it's a no brainer to

73
00:04:15.159 --> 00:04:15.879
<v Speaker 1>show them stuff.

74
00:04:16.040 --> 00:04:17.600
<v Speaker 2>Anything they see they can buy.

75
00:04:18.360 --> 00:04:21.800
<v Speaker 1>So it is a it's it's the wealthiest advertising demographic

76
00:04:22.319 --> 00:04:27.439
<v Speaker 1>I think of any platform social media, orial media on Earth,

77
00:04:28.279 --> 00:04:32.040
<v Speaker 1>and I really want that message to resonate. Something that

78
00:04:32.079 --> 00:04:35.759
<v Speaker 1>I think is actually very important is is to actually

79
00:04:35.920 --> 00:04:39.720
<v Speaker 1>let the let the grock API, let the grock AI

80
00:04:40.160 --> 00:04:44.879
<v Speaker 1>actually figure out the right targeting. Sometimes advertisers want to say, okay,

81
00:04:44.920 --> 00:04:47.600
<v Speaker 1>we want to target men or women of a particular

82
00:04:47.639 --> 00:04:51.240
<v Speaker 1>age or something like that. But but actually, if you're

83
00:04:51.639 --> 00:04:54.800
<v Speaker 1>i mean, for augment's sake, let's say you're selling gardening tools,

84
00:04:54.839 --> 00:04:57.759
<v Speaker 1>you just care that somebody likes gardening. You don't care

85
00:04:57.839 --> 00:05:00.480
<v Speaker 1>if they're an eighty year old woman or twenty year

86
00:05:00.480 --> 00:05:03.279
<v Speaker 1>old men that you just care that they they're interested

87
00:05:03.279 --> 00:05:04.399
<v Speaker 1>in gardening.

88
00:05:04.879 --> 00:05:06.560
<v Speaker 2>And that's what groc can figure out.

89
00:05:07.040 --> 00:05:10.959
<v Speaker 1>So historically you have to do these yesses based on

90
00:05:11.040 --> 00:05:13.839
<v Speaker 1>demographic breakdown, but you don't have to do that anymore.

91
00:05:13.839 --> 00:05:16.439
<v Speaker 1>You can literally just let the AI do everything and

92
00:05:16.480 --> 00:05:17.959
<v Speaker 1>it actually ends up with a better outcome.

93
00:05:18.480 --> 00:05:22.439
<v Speaker 2>So that's I'd encourage, you know, just just let the AI.

94
00:05:22.360 --> 00:05:26.439
<v Speaker 1>Do its thing and and it'll it'll find your product,

95
00:05:26.439 --> 00:05:29.360
<v Speaker 1>it will match the product service to the user's needs,

96
00:05:29.439 --> 00:05:33.279
<v Speaker 1>and and that'll actually be the best best outcome if

97
00:05:33.360 --> 00:05:37.079
<v Speaker 1>you advertise a part to service that the user needs

98
00:05:37.720 --> 00:05:40.959
<v Speaker 1>when they want to, that's the ideal moment. So like,

99
00:05:41.120 --> 00:05:43.720
<v Speaker 1>let's say, in speaking of reply ads, that there's a

100
00:05:43.720 --> 00:05:48.639
<v Speaker 1>reply thread discussing auto insurance and someone's maybe unhappy with

101
00:05:48.680 --> 00:05:53.120
<v Speaker 1>their existing auto insurance provider, and they're discussing, you know what,

102
00:05:53.120 --> 00:05:55.399
<v Speaker 1>what what the other options are? That would be the

103
00:05:55.439 --> 00:06:01.079
<v Speaker 1>perfect location to put an car insurance. It's like, what

104
00:06:01.160 --> 00:06:06.720
<v Speaker 1>about us? Because that's that's that's both what somebody is interested.

105
00:06:06.399 --> 00:06:09.279
<v Speaker 2>In and when they're interested in it. So reply ads

106
00:06:09.279 --> 00:06:10.240
<v Speaker 2>can be very powerful.

107
00:06:11.160 --> 00:06:13.959
<v Speaker 1>Talking really just means reducing the probability that an AD

108
00:06:14.160 --> 00:06:16.519
<v Speaker 1>is spam for a user, then an AD is not applicable.

109
00:06:16.959 --> 00:06:19.680
<v Speaker 1>We do actually care about the aesthetics of ads. That

110
00:06:19.800 --> 00:06:23.920
<v Speaker 1>a beautiful AD is actually quite important to us because

111
00:06:24.240 --> 00:06:27.839
<v Speaker 1>if an AD is ugly or strident, that damages our product.

112
00:06:28.279 --> 00:06:30.560
<v Speaker 1>If somebody's scrolling through and they see an AD and

113
00:06:30.600 --> 00:06:34.360
<v Speaker 1>that that ad is jarring, and and you know, just

114
00:06:34.439 --> 00:06:37.920
<v Speaker 1>really upsetting or ugly, then that actually damages our user

115
00:06:37.959 --> 00:06:41.680
<v Speaker 1>experience and that damages our products. So we actually want

116
00:06:41.680 --> 00:06:46.720
<v Speaker 1>to favor ads that are beautiful. So that just just

117
00:06:46.920 --> 00:06:49.639
<v Speaker 1>clear that that that ads will will receive, you know,

118
00:06:49.680 --> 00:06:53.439
<v Speaker 1>a slight boost if they are aesthetically I mean, we're

119
00:06:53.480 --> 00:06:56.120
<v Speaker 1>not talking about the need to win some contests or something,

120
00:06:56.240 --> 00:06:59.040
<v Speaker 1>just just not jarring and ugly. So I have a

121
00:06:59.079 --> 00:07:02.279
<v Speaker 1>personal aversion to are and ugly ads, and I'd actually

122
00:07:02.360 --> 00:07:03.240
<v Speaker 1>rather make less money.

123
00:07:03.439 --> 00:07:05.480
<v Speaker 2>Then if then have you.

124
00:07:05.439 --> 00:07:08.399
<v Speaker 1>Know, strident and ugly ads in the system, and then

125
00:07:08.439 --> 00:07:11.639
<v Speaker 1>we are changing the cost of an ad proportionate to

126
00:07:11.800 --> 00:07:14.839
<v Speaker 1>how much vertical space it takes, just as a full

127
00:07:14.879 --> 00:07:17.759
<v Speaker 1>page ad in a newspaper costs a lot more than

128
00:07:18.040 --> 00:07:21.560
<v Speaker 1>a quarter page ad. The smaller the ad is, the

129
00:07:21.600 --> 00:07:24.519
<v Speaker 1>more ad impressions will get for a given dollar. But

130
00:07:24.600 --> 00:07:26.120
<v Speaker 1>even in that case, so I think, you know, there's

131
00:07:26.120 --> 00:07:29.040
<v Speaker 1>could be a lot of blank space around the ad itself.

132
00:07:29.399 --> 00:07:32.959
<v Speaker 1>It should really be the image should be shortened to

133
00:07:33.000 --> 00:07:34.639
<v Speaker 1>be relevant. There's no point in having a bunch of

134
00:07:34.680 --> 00:07:37.800
<v Speaker 1>blank space around that we wanted. General brand advertising can

135
00:07:37.800 --> 00:07:40.720
<v Speaker 1>certainly find, but that doesn't require much targeting if it's

136
00:07:40.759 --> 00:07:44.600
<v Speaker 1>a big brand. But for specific products, Like so your product,

137
00:07:44.720 --> 00:07:47.480
<v Speaker 1>you know, helps people sleep better, and that obviously would

138
00:07:47.560 --> 00:07:51.519
<v Speaker 1>be the ideal time to show users and add for

139
00:07:51.560 --> 00:07:54.279
<v Speaker 1>your product. It would be if there's any discussion about sleep,

140
00:07:54.360 --> 00:07:56.879
<v Speaker 1>or if they've made some comment about like not getting

141
00:07:56.959 --> 00:07:59.279
<v Speaker 1>enough good sleep and maybe the a sleep you know

142
00:07:59.360 --> 00:08:01.920
<v Speaker 1>mattress would solve their their issues.

143
00:08:02.319 --> 00:08:03.839
<v Speaker 2>That's that's obviously would be perfect.

144
00:08:03.920 --> 00:08:08.319
<v Speaker 1>And then you know, I'm particularly looking for unregretted user purchases.

145
00:08:08.480 --> 00:08:11.519
<v Speaker 1>So a user sees a product or service, buys a

146
00:08:11.519 --> 00:08:13.800
<v Speaker 1>product or service and it's and it solves their need,

147
00:08:14.240 --> 00:08:16.120
<v Speaker 1>then I think we've done something fundamentally good.

148
00:08:16.199 --> 00:08:17.120
<v Speaker 2>Yeah. Yeah.

149
00:08:17.680 --> 00:08:20.319
<v Speaker 1>If if people have received a product or service that

150
00:08:20.360 --> 00:08:23.519
<v Speaker 1>has made their lives better so even that they have

151
00:08:23.600 --> 00:08:27.120
<v Speaker 1>no post post purchase regret, in fact, they're happy, then

152
00:08:27.120 --> 00:08:29.279
<v Speaker 1>we've done a good thing for the world. So that

153
00:08:29.279 --> 00:08:32.399
<v Speaker 1>that's our goal. I'd really advise to be as specific

154
00:08:32.480 --> 00:08:35.159
<v Speaker 1>as possible. So even if you've got say a large

155
00:08:35.200 --> 00:08:39.759
<v Speaker 1>number of products, my recommendation would be to put what

156
00:08:39.879 --> 00:08:43.600
<v Speaker 1>you think are the most likely products that people would buy,

157
00:08:43.679 --> 00:08:45.399
<v Speaker 1>because if you give sometimes if you give people too

158
00:08:45.480 --> 00:08:47.360
<v Speaker 1>much choice, they're like, I don't know what to do here.

159
00:08:47.840 --> 00:08:50.279
<v Speaker 1>You know, maybe just give them what is probably the

160
00:08:50.279 --> 00:08:53.879
<v Speaker 1>best choice and then they can always you know, maybe

161
00:08:54.240 --> 00:08:56.039
<v Speaker 1>go to the website and change that or something.

162
00:08:56.039 --> 00:08:57.679
<v Speaker 2>But if you try to give.

163
00:08:57.600 --> 00:09:01.720
<v Speaker 1>People as as specific as solution to the problem as possible,

164
00:09:01.759 --> 00:09:03.759
<v Speaker 1>I think that's the right way to go. Yeah, and

165
00:09:03.799 --> 00:09:06.519
<v Speaker 1>you can, you can upload several several products, but but

166
00:09:06.559 --> 00:09:10.519
<v Speaker 1>it's you know, and it's I think it's like sometimes

167
00:09:10.360 --> 00:09:14.320
<v Speaker 1>the product carousel. That's okay, but you really you really

168
00:09:14.360 --> 00:09:16.519
<v Speaker 1>want to just you only have the user intention for

169
00:09:16.679 --> 00:09:18.639
<v Speaker 1>maybe a second or two, and you want to just

170
00:09:18.679 --> 00:09:21.399
<v Speaker 1>have that. It's like, okay, it's this one, go for it.

171
00:09:21.399 --> 00:09:23.639
<v Speaker 1>And you can put in different products and see which

172
00:09:23.679 --> 00:09:26.120
<v Speaker 1>ones are more resonant. But we're trying to be as

173
00:09:26.279 --> 00:09:30.600
<v Speaker 1>as precise as possible, so so we have the highest

174
00:09:30.639 --> 00:09:33.200
<v Speaker 1>product that grock has the highest probability of matching a

175
00:09:33.279 --> 00:09:36.360
<v Speaker 1>product of service with product product or service with the

176
00:09:36.480 --> 00:09:39.399
<v Speaker 1>user's actual needs. Yeah, in fact, you can do you

177
00:09:39.440 --> 00:09:41.159
<v Speaker 1>can do that right now if you if you if

178
00:09:41.159 --> 00:09:45.279
<v Speaker 1>you use grock and you can actually ask grock just

179
00:09:45.360 --> 00:09:48.519
<v Speaker 1>go you know, maybe use grock expert Mode. And I'd

180
00:09:48.559 --> 00:09:50.679
<v Speaker 1>recommend this to everyone because you can do it right now.

181
00:09:51.120 --> 00:09:55.399
<v Speaker 1>Is you know, maybe you can post a probably the

182
00:09:55.440 --> 00:09:59.039
<v Speaker 1>best way to do this, certainly for copy, for text copy,

183
00:09:59.080 --> 00:10:01.480
<v Speaker 1>you could say what is the best way to represent

184
00:10:01.519 --> 00:10:04.000
<v Speaker 1>this and and ask GROC to give you some recommendations.

185
00:10:04.279 --> 00:10:06.279
<v Speaker 1>You can literally just do that on groc and the

186
00:10:06.440 --> 00:10:09.000
<v Speaker 1>X app or the groc app or grock dot com

187
00:10:09.840 --> 00:10:12.840
<v Speaker 1>and I think you'll see some interesting suggestions because grock.

188
00:10:12.960 --> 00:10:16.279
<v Speaker 1>Actually one of the big advantages of the GROC AI

189
00:10:16.879 --> 00:10:21.759
<v Speaker 1>is that it's connected to the the the X system,

190
00:10:21.919 --> 00:10:24.519
<v Speaker 1>so it actually understands the X system. It's ingested the

191
00:10:24.639 --> 00:10:27.720
<v Speaker 1>entire it's done training on the entire X you know,

192
00:10:27.720 --> 00:10:30.279
<v Speaker 1>all of the public user posts it's done training on,

193
00:10:30.720 --> 00:10:32.960
<v Speaker 1>so it actually has a good a good sense of

194
00:10:33.000 --> 00:10:35.960
<v Speaker 1>what people respond to already. You just literally asked Rock

195
00:10:36.039 --> 00:10:39.600
<v Speaker 1>right now. Yeah, I mean, so we have sort of

196
00:10:39.600 --> 00:10:42.399
<v Speaker 1>what I call grock, you know, grock. Imagine it's kind

197
00:10:42.399 --> 00:10:45.440
<v Speaker 1>of version zero point one, kind of beta. And so

198
00:10:45.639 --> 00:10:49.279
<v Speaker 1>you can take any image and and and turn that

199
00:10:49.320 --> 00:10:52.279
<v Speaker 1>into a six second video with the with rock right

200
00:10:52.320 --> 00:10:55.279
<v Speaker 1>now in like sixteen seconds, and you can you can

201
00:10:55.399 --> 00:10:58.679
<v Speaker 1>you can grock will it's somewhat will randomize it, so

202
00:10:58.720 --> 00:11:03.080
<v Speaker 1>you can actually do manyferent videos and generally the video

203
00:11:03.080 --> 00:11:05.120
<v Speaker 1>is going to be a little bit more compelling than

204
00:11:05.440 --> 00:11:08.360
<v Speaker 1>a static image, and you can also give rock com

205
00:11:08.440 --> 00:11:12.039
<v Speaker 1>matching custom instructions. This is just kind of I literally

206
00:11:12.080 --> 00:11:14.159
<v Speaker 1>call it boson zero point one of the our product.

207
00:11:14.200 --> 00:11:17.840
<v Speaker 1>We've got a much more sophisticated version coming out in

208
00:11:17.919 --> 00:11:20.240
<v Speaker 1>two or three weeks where you can which way you

209
00:11:20.279 --> 00:11:24.519
<v Speaker 1>can add voiceover, you can add in fact, technically right

210
00:11:24.559 --> 00:11:26.240
<v Speaker 1>now we just added a future where you can do

211
00:11:26.320 --> 00:11:29.440
<v Speaker 1>some voiceover, so it's worth trying out.

212
00:11:29.600 --> 00:11:30.600
<v Speaker 2>Like basically, I check.

213
00:11:31.960 --> 00:11:34.200
<v Speaker 1>Update your grock app and your AX app almost every

214
00:11:34.240 --> 00:11:37.279
<v Speaker 1>day because we're rolling out upgrades and features and bug

215
00:11:37.279 --> 00:11:41.120
<v Speaker 1>fixes almost every day. And yeah, you'll just be able

216
00:11:41.159 --> 00:11:43.559
<v Speaker 1>to say that Grock just make me a video ad

217
00:11:43.639 --> 00:11:46.240
<v Speaker 1>or an image ad or whatever or text ad that

218
00:11:46.440 --> 00:11:49.440
<v Speaker 1>is that accurately describes my product and it is likely

219
00:11:49.480 --> 00:11:52.159
<v Speaker 1>to appeal to the audience based on what Grock knows

220
00:11:52.159 --> 00:11:54.679
<v Speaker 1>about X, and it pretty much knows everything about X,

221
00:11:54.799 --> 00:11:58.759
<v Speaker 1>then yeah, it should be super easy to create a

222
00:11:58.799 --> 00:12:02.080
<v Speaker 1>text image or video ad. You can upload any image

223
00:12:02.159 --> 00:12:04.399
<v Speaker 1>right now, including an image of your product, to two grock,

224
00:12:04.480 --> 00:12:06.240
<v Speaker 1>so you can do it from your phone in like

225
00:12:06.279 --> 00:12:10.759
<v Speaker 1>five seconds, so any or or from your browser you

226
00:12:10.799 --> 00:12:12.559
<v Speaker 1>can take You can take an image of your product

227
00:12:12.720 --> 00:12:15.440
<v Speaker 1>and turn that into a video and then you can

228
00:12:15.480 --> 00:12:19.240
<v Speaker 1>actually say, now add something new to the video. But

229
00:12:19.320 --> 00:12:22.279
<v Speaker 1>things are evolving very quickly, so like you said, up

230
00:12:22.360 --> 00:12:26.240
<v Speaker 1>they trap almost every day and there'll be new features

231
00:12:26.240 --> 00:12:29.519
<v Speaker 1>and functionality. The rate of evolution is very fast, and

232
00:12:29.919 --> 00:12:33.159
<v Speaker 1>we are implementing that. We are seeing a little bit

233
00:12:33.200 --> 00:12:35.440
<v Speaker 1>of where the same ad is shown to too many people,

234
00:12:35.559 --> 00:12:38.360
<v Speaker 1>the over frequency issue. In fact that that was just

235
00:12:38.440 --> 00:12:44.080
<v Speaker 1>raising that with Roman Uh right before this meeting, and

236
00:12:44.240 --> 00:12:47.200
<v Speaker 1>we should see a significant proven to that by Friday.

237
00:12:47.240 --> 00:12:50.919
<v Speaker 1>We're actually pushing a change that reduces the over frequency

238
00:12:50.960 --> 00:12:54.480
<v Speaker 1>problem literally in two days. This's clearly times when a

239
00:12:54.519 --> 00:12:57.159
<v Speaker 1>single users shown the same ad far too many times,

240
00:12:57.159 --> 00:12:59.440
<v Speaker 1>Like it's so you know, one hundred times type of thing,

241
00:12:59.480 --> 00:13:04.240
<v Speaker 1>and that's that's silly. So we're going to iterate until

242
00:13:04.279 --> 00:13:06.960
<v Speaker 1>we see, like what what actually results in an effective

243
00:13:07.519 --> 00:13:10.799
<v Speaker 1>and purchase experience. You know, past a certain point, you're

244
00:13:11.120 --> 00:13:14.159
<v Speaker 1>really just you know, being a dead horse.

245
00:13:14.159 --> 00:13:15.000
<v Speaker 2>It's kind of pointless.

246
00:13:15.240 --> 00:13:19.320
<v Speaker 1>I guess generally when any product breaks with especially AI,

247
00:13:19.639 --> 00:13:21.919
<v Speaker 1>but really almost any product or any kind of new thing,

248
00:13:22.279 --> 00:13:24.879
<v Speaker 1>if it's it's going to trend on X almost immediately,

249
00:13:25.639 --> 00:13:28.159
<v Speaker 1>so you can kind of tell if a product is

250
00:13:28.159 --> 00:13:30.159
<v Speaker 1>going to be successful actually because it's going to trend

251
00:13:30.279 --> 00:13:32.240
<v Speaker 1>organically on X in a big way.

252
00:13:32.679 --> 00:13:34.120
<v Speaker 2>And I guess we would also see it with.

253
00:13:34.279 --> 00:13:37.440
<v Speaker 1>Rock, you know, when people ask Rock questions, you can

254
00:13:37.480 --> 00:13:40.360
<v Speaker 1>see that frequency spike when there's something new and interesting.

255
00:13:41.000 --> 00:13:45.000
<v Speaker 1>We do think that immersive video ad advertising is important

256
00:13:46.039 --> 00:13:49.240
<v Speaker 1>and that that will be a significant focus in the future.

257
00:13:49.600 --> 00:13:52.320
<v Speaker 1>So if you're in immersive video, you should see a

258
00:13:52.360 --> 00:13:55.840
<v Speaker 1>massive compelling video ads. And then the same thing is

259
00:13:55.879 --> 00:13:59.679
<v Speaker 1>true with when when people do search with l MS,

260
00:14:00.000 --> 00:14:02.559
<v Speaker 1>at least with GROC, which we should show people products

261
00:14:02.600 --> 00:14:05.039
<v Speaker 1>that they would find interesting. If they're a false famous

262
00:14:05.039 --> 00:14:07.440
<v Speaker 1>manner in our system, we will that that ad will

263
00:14:07.440 --> 00:14:09.120
<v Speaker 1>be struck from the system.

264
00:14:09.360 --> 00:14:11.559
<v Speaker 2>That it would it would it would work well.

265
00:14:11.600 --> 00:14:15.519
<v Speaker 1>I don't know about nuances, but if if, if misleading

266
00:14:15.519 --> 00:14:17.559
<v Speaker 1>claims are made, then the then that ad will be

267
00:14:17.559 --> 00:14:19.679
<v Speaker 1>struck from the system and that advertiser will be banned

268
00:14:19.679 --> 00:14:23.039
<v Speaker 1>from X. We actually have GROC assessed the scamminess of

269
00:14:23.039 --> 00:14:25.879
<v Speaker 1>an AD, so there are far fewer scamming ads that

270
00:14:25.960 --> 00:14:28.200
<v Speaker 1>actually come through because Grock's pretty smart and GROK gets

271
00:14:28.200 --> 00:14:29.039
<v Speaker 1>smarter over time.

272
00:14:29.600 --> 00:14:32.080
<v Speaker 2>So if an AD is misleading.

273
00:14:32.480 --> 00:14:35.840
<v Speaker 1>You know, GROC I think is going to figure that out, yes, right,

274
00:14:35.960 --> 00:14:38.639
<v Speaker 1>telling ads without being misleading. There have been a lot

275
00:14:38.679 --> 00:14:41.879
<v Speaker 1>of improvements behind the scenes, So admittedly the interface is suboptimal,

276
00:14:42.000 --> 00:14:44.080
<v Speaker 1>but in terms of matching, there's been there have been

277
00:14:44.080 --> 00:14:47.720
<v Speaker 1>significant improvements behind the scenes, and we have seen advertiseacy,

278
00:14:48.080 --> 00:14:51.639
<v Speaker 1>you know, double or triple the conversion rates that they

279
00:14:52.000 --> 00:14:55.320
<v Speaker 1>so we're seeing even a few months ago. So it

280
00:14:55.360 --> 00:14:57.840
<v Speaker 1>may not on the surface appear like there's a lot happening,

281
00:14:57.879 --> 00:15:00.039
<v Speaker 1>but there's certainly a lot happening beneath the surface. I

282
00:15:00.039 --> 00:15:02.399
<v Speaker 1>don't believe in disruption for the sake of disruption, and

283
00:15:02.440 --> 00:15:04.960
<v Speaker 1>I mean, frankly, for the X system.

284
00:15:05.120 --> 00:15:06.200
<v Speaker 2>We're just trying to get.

285
00:15:06.240 --> 00:15:10.440
<v Speaker 1>To where we were at a level of capability that

286
00:15:10.559 --> 00:15:13.879
<v Speaker 1>is kind of industry standard. The sort of the old

287
00:15:13.919 --> 00:15:16.559
<v Speaker 1>Twitter advertising system basically had no targeting, so it was

288
00:15:16.639 --> 00:15:20.279
<v Speaker 1>just kind of random spam, and we're trying to get

289
00:15:20.320 --> 00:15:23.279
<v Speaker 1>to and we're making progress rapidly to the point where

290
00:15:23.600 --> 00:15:26.720
<v Speaker 1>we're showing our users products and services that they actually

291
00:15:26.720 --> 00:15:29.639
<v Speaker 1>want to buy. So we're coming from a kind of

292
00:15:29.639 --> 00:15:34.000
<v Speaker 1>a tough position but making rapid progress to showing users

293
00:15:34.120 --> 00:15:37.039
<v Speaker 1>products and services that they actually want to buy the

294
00:15:37.080 --> 00:15:39.200
<v Speaker 1>acid test, being that you're on the X system. If

295
00:15:39.200 --> 00:15:41.000
<v Speaker 1>you're on the X system for several hours a week,

296
00:15:41.120 --> 00:15:44.320
<v Speaker 1>you should have probably bought something or certainly seen as

297
00:15:44.360 --> 00:15:46.720
<v Speaker 1>for lots of things that you'd want to buy. We're

298
00:15:46.720 --> 00:15:51.399
<v Speaker 1>trying not to inject editorial into advertising apart from the

299
00:15:51.399 --> 00:15:54.200
<v Speaker 1>fact that we want ads to be esthetically pleasing, to

300
00:15:54.240 --> 00:15:57.879
<v Speaker 1>be beautiful, and not to be shrill in any way,

301
00:15:58.399 --> 00:16:01.879
<v Speaker 1>and we want them to be accurate. That's that's and

302
00:16:01.600 --> 00:16:04.440
<v Speaker 1>and matched to a user's needs.

303
00:16:04.480 --> 00:16:05.440
<v Speaker 2>That's it.

304
00:16:05.519 --> 00:16:08.600
<v Speaker 1>So Yeah, to be to be clear, like, we're not

305
00:16:08.639 --> 00:16:11.480
<v Speaker 1>saying that an AD needs to be winning most Beautiful

306
00:16:11.480 --> 00:16:14.720
<v Speaker 1>out of the Year award, but but that it just

307
00:16:14.799 --> 00:16:17.039
<v Speaker 1>can't be extremely ugly and strident.

308
00:16:17.919 --> 00:16:18.399
<v Speaker 2>That's all.

309
00:16:18.799 --> 00:16:22.279
<v Speaker 1>So if if Grock determines that an AD is you know,

310
00:16:22.840 --> 00:16:27.240
<v Speaker 1>just shrolls strident, ugly and discordant, it will receive a penalty.

311
00:16:28.120 --> 00:16:31.759
<v Speaker 1>But it really an okay AD will not receive a penalty.

312
00:16:31.960 --> 00:16:35.360
<v Speaker 1>It doesn't need to be winning any you know, major

313
00:16:35.399 --> 00:16:39.519
<v Speaker 1>aesthetic contest. It just needs to not be aggravating. Yeah, absolutely,

314
00:16:39.600 --> 00:16:43.360
<v Speaker 1>Grock will do this dynamically. So this should just happen automatically.

315
00:16:43.440 --> 00:16:46.919
<v Speaker 1>Just put the ad in and it should react automatically

316
00:16:46.960 --> 00:16:49.000
<v Speaker 1>to any anything that's happening in real time.

317
00:16:49.639 --> 00:16:52.279
<v Speaker 2>So and with that, yeah, thank you for tuning in.

318
00:16:52.440 --> 00:16:54.159
<v Speaker 2>We're working hard to make x.

319
00:16:54.240 --> 00:16:58.679
<v Speaker 1>A a great platform for advertising, ultimately the best advertising,

320
00:16:58.960 --> 00:17:02.360
<v Speaker 1>the best platform for apazing and appreciate your support.

321
00:17:02.440 --> 00:17:02.720
<v Speaker 2>Thank you,