WEBVTT

1
00:00:00.100 --> 00:00:17.960
[upbeat music] Welcome to the Rebooting Show. I am Brian Morrissey. The rise of generative AI is undoing one of the web's foundational bargains.

2
00:00:18.280 --> 00:00:29.860
For years, publishers let Google crawl their content, they got traffic in return, and they monetized it mostly with ads running through Google's pipes. It was an imperfect bargain, but at least it was fairly clear.

3
00:00:30.580 --> 00:00:36.020
Google set the rules, and you followed them, and if you executed on them as a publisher, you had a business.

4
00:00:36.760 --> 00:00:48.140
But now, with the rollout of AI-generated summaries and answers with AI overviews, that exchange is just breaking down. In fact, it's being scrapped altogether.

5
00:00:48.180 --> 00:00:58.240
I mean, just look at how last week Google rolled out a Offer Walls, they're calling it, that lets publishers monetize traffic better, and this is a classic case of too little too late.

6
00:00:58.300 --> 00:01:08.320
And it happens to come as publishers of all sizes are seeing market traffic declines from Google that everyone with eyes and a few working brain cells believes will only accelerate.

7
00:01:08.380 --> 00:01:13.940
The prospect of Google Zero has gone from some kind of worst-case scenario planning to reality.

8
00:01:14.570 --> 00:01:40.760
At Cannes this year, incongruously outside of a villa and next to a pool, I spoke with Neil Vogel, the CEO of Dotdash Meredith, about how they're preparing for a world where Google sends little to no traffic, and, and he was very clear-eyed that this isn't about fighting change, it's about being ready for it, and they are leaning into product, brand, and direct engagement because they know the economics of the open web are shifting under their feet.

9
00:01:40.890 --> 00:01:53.180
And so every publisher I talk to, this is, you know, just reality. The old system simply isn't coming back, and there is an obvious impulse, and I understand it, to mourn that and to point fingers at culprits.

10
00:01:53.260 --> 00:02:00.580
I have many to, to point fingers at. Maybe I'll do that later this summer. But none of this strikes me as particularly, you know, useful. It, it...

11
00:02:00.640 --> 00:02:06.080
Focusing on fairness, I find, makes you feel good, but it doesn't really do much in reality.

12
00:02:06.180 --> 00:02:17.660
You know, I think I see a lot of impatience to sort of move on because, you know, nobody's coming to save, you know, publishing, I don't think, and it will need to find a way out on its own. I think it will.

13
00:02:17.900 --> 00:02:28.920
I think it's just gonna be messy. And that's just because, you know, when it comes to, to these AI engines, the content has already been ingested, these models have been trained, the value has been stripped.

14
00:02:29.340 --> 00:02:41.180
There's little chance that I see, just based on my experience of having covered this industry for a generation, that Big Tech is going to all of a sudden offer retroactive compensation out of some kind of principle.

15
00:02:41.300 --> 00:02:47.140
Just simply not happening. The real question is what comes next? What would a new economic bargain look like?

16
00:02:47.220 --> 00:02:56.120
One that, that hopefully gives publisher, publishers a stake in the systems that are profiting off of their work. You know, I, I think that has to start with attribution.

17
00:02:56.300 --> 00:03:01.160
I mean, I listen to many podcasts from tech leaders, and I listen to, I listen to the words they, they use.

18
00:03:01.720 --> 00:03:10.000
A-and what's clear is what they are doing, at least it's clear to me, is another instance of obfuscation by positioning the rampant theft as quote-unquote, "training."

19
00:03:10.600 --> 00:03:19.500
Unlike with Napster, they made sure that the outputs are not carbon copies, and therefore, you know, any claims of copyright are irrelevant.

20
00:03:19.530 --> 00:03:28.800
It's the classic case of the plagiarist who reorders the words and combines many sources and then pretends that the work that they have created is quote-unquote, "original."

21
00:03:29.020 --> 00:03:39.200
I have little faith in the courts, particularly following the recent ruling in favor of Anthropic's AI training on copyrighted books. So the question is what comes next?

22
00:03:39.300 --> 00:03:50.300
In this episode, I speak with Annalise Janssen, who is the chief business officer at ProRata, about how publishers can respond in this moment, 'cause it's an important one. ProRata is a startup founded by Bill Gross.

23
00:03:50.520 --> 00:04:00.840
If you don't know Bill Gross, you should, because he is the original architect of paid search. He basically invented it. Google copied it. And I don't say that, like, out of my judgment, it's the judgment of a court.

24
00:04:00.960 --> 00:04:08.780
Google had to pay Yahoo, which had acquired Overture, which was the company that Bill had started, and paid a settlement because of that.

25
00:04:09.480 --> 00:04:22.400
So what Bill and ProRata are doing now is trying to build a new infrastructure for basically ethical AI engagement, and the idea really comes down to attribution, and that is if you can count it, you can compensate for it.

26
00:04:22.480 --> 00:04:38.240
You know, attribution becomes the foundation for this new economic model, one that pays publishers not just for clicks but for their contributions to the, the content that is being produced by these AI engines and that they are profiting from.

27
00:04:38.770 --> 00:04:44.160
And now, how that looks is gonna be messy. I mean, this is something I had spoken to Neil about.

28
00:04:44.240 --> 00:04:55.300
I mean, in his view, you know, you need to get leverage in these negotiations because, you know, publishers have never been able to exert themselves for a whole variety of reasons.

29
00:04:55.340 --> 00:05:01.560
You know, I think a lot of times people compare the publishing industry to, you know, for instance, music, right?

30
00:05:01.650 --> 00:05:09.320
'Cause if you go on these AI engines, I'm always struck, try to get the lyrics of a song out of one of these AI engines, and you cannot, okay?

31
00:05:09.740 --> 00:05:22.300
And that's because it's an oligopoly in music, and publishing is incredibly fragmented, and these tech companies want to fragment it even further, and that's why they're all into creators, and I could go on about that ad nauseam, but I won't.

32
00:05:22.400 --> 00:05:39.680
So I think when you think about what a new bargain would look like, it, it could mean, i- you know, embedding attributions into these AI-generated answers and creating frameworks for a pay-per-use compensation and, and giving media companies tools themselves to build AI-driven products.

33
00:05:39.740 --> 00:05:51.560
I mean, if AI is gonna become the new distribution layer, then publishing is gonna need a new monetization layer that matches it and, and works well with it. I mean, otherwise, I think the risk is obvious.

34
00:05:51.620 --> 00:06:11.096
I mean, quality content will become simply an economically unsustainable activity.So Annelies and I talk about what such a fair system would look like and why the current dynamics echoes past failures to organize collectively, and also whether publishers will ultimately become, in effect, wholesalers to these answer engines.

35
00:06:11.136 --> 00:06:18.796
This is something that I've been going on about, like, a bit just because it, it seems pretty clear to me that that is one path that this could take.

36
00:06:19.416 --> 00:06:30.696
This is overall a conversation about what happens, you know, after traffic, after the eyeballs here, and how, you know, the value of quality content can still be captured by those who create it and also how to get there.

37
00:06:31.256 --> 00:06:38.516
As always, love to hear your feedback. My email is brian@therebooting.com. Now here's my conversation with Annelies.

38
00:06:41.076 --> 00:06:50.316
[upbeat music] I did a little podcast, we're trying this with Ben Smith and Peter Kafka where, you know, to...

39
00:06:50.386 --> 00:06:59.516
'cause podcast, like, discovery is horrific, so we figure we have overlapping audiences, but we're kinda different. We're different. So we're gonna, like, split it up between, like, the three.

40
00:07:00.056 --> 00:07:10.216
And then I find out on the podcast that Ben fakes the intro. Like, it sounds like he's like, you know, he acts like he's like, "Okay," and now we'll, like, you know, bring him, but he does that after the podcast.

41
00:07:10.296 --> 00:07:21.556
He's like, "It's show business." I'm like, "I don't do that." [laughs] Yeah, because you can highlight, you know, you can take specific things that you've heard th- through the podcast. I think there's a word for it.

42
00:07:21.616 --> 00:07:32.016
Is it, is it podcast frenemies? Po- frenemies? Yeah. Pod- pod- Frenemies. Isn't it? Pod enemies, as in, you know, make sure that you appear on your neighbor's podcast. Yeah.

43
00:07:32.056 --> 00:07:43.556
So I'm gonna do the intro later just to make sure I don't butcher your surname. Is it, is it Janssen? Janssen? Janssen. Uh, I listen to many pronunciations from my- I know, but which do you prefer? My first name.

44
00:07:43.716 --> 00:07:52.736
I, I think Annelies Janssen. Janssen is fine. Okay. But it's really not. It's... Well, if you were Dutch and you would speak to me in Dutch, you would say Janssen. [speaking Dutch] Yeah.

45
00:07:52.876 --> 00:08:10.436
So you would say Jan- [laughs] You'd say Janssen. Janssen. Yeah. Okay. [speaking Dutch] I don't know. I, I lived in, um, in Leuven for two years. Oh, wow. To study- Yeah... or? Fantastic. Studying.

46
00:08:10.496 --> 00:08:26.045
My parents lived in a Flemish commune called Overijse just outside of Brussels, and I, I went to a year of college there, and then I went to grad school. Wow. Yeah. So you learned how to drink beer. [laughs] Basically.

47
00:08:26.336 --> 00:08:41.456
Oud Markt, that was, that was my degree was in- Yeah... Oud Markt studies. [laughs] There's a lot to learn when you drink beer. Annelies, thank you for joining me to talk about AI and impending doom. Wow. Yes.

48
00:08:41.696 --> 00:08:49.096
[laughs] It's, it depends on how you look at it. I know. I know. But I wanna start with, we're gonna get into pro rata and just...

49
00:08:49.196 --> 00:09:02.676
But I wanna start with something that we have discussed via email and before this podcast, but I almost feel like this is one where publishers are not actually taking the, the threat as seriously as they should.

50
00:09:02.756 --> 00:09:13.336
Like, what is happening on many fronts with AI, but most particularly with traffic, I almost feel like people are not taking it as seriously as they should.

51
00:09:14.196 --> 00:09:19.605
Just give me, I wanna do a little exercise, give me a worst-case scenario in five years. Okay.

52
00:09:19.656 --> 00:09:27.516
With where you see things are going, you've got experiences on all side of, of this, but, like, just to sort of paint a picture, what's a worst-case scenario?

53
00:09:27.556 --> 00:09:36.226
I'm not saying that's gonna happen, and then we'll get to a best-case scenario. Well, thank you for this, and it's a real, it's a real pleasure to be in this podcast and to continue our conversation, Brian.

54
00:09:36.876 --> 00:09:46.336
Let me start with something that I see is happening within the industry. I think the big steal is being normalized. I personally find that very concerning.

55
00:09:46.416 --> 00:09:58.716
I think a year ago people were much more aware that their content was sto- I think a year ago there was l- a, a lot more discussion and people speaking up about content being stolen.

56
00:09:58.946 --> 00:10:11.056
I don't see that anymore, or less, way less of it. I was just attending a number of conferences where it's j- not the topic of conversation. But let's go back to your question. So where do I see this play out?

57
00:10:11.096 --> 00:10:27.436
And obviously, you know, and I am biased, um, I have a playout without pro rata, or as one could call it, a, a product solution for the whole industry and not just the happy few who are able to do a deal, and, and a product solution with pro rata.

58
00:10:28.376 --> 00:10:41.476
But the big steal is happening. All content's taken. And share of wallet, share of voice, share of audience is with those, and already was with those who did deals, who have been lucky enough to do, to do deals.

59
00:10:42.046 --> 00:10:53.086
And the deals, whether they are licensing deals or commercial deals, they're get out of the courtroom deals, I call them. And those w- Get out of the boardroom deals? What is it? Get, get out of the courtroom deals.

60
00:10:53.156 --> 00:10:56.376
Oh, courtroom. Okay. Yeah. They're payoffs. Yes.

61
00:10:56.496 --> 00:11:15.816
And so if we continue to live in this bifurcated media world where we have the happy view 0.5% with some sort of money based on, you know, you called it licensing, I don't think it's licensing, depends, but I'm not a lawyer, and the ones who, who don't, the haves and the have-not.

62
00:11:16.436 --> 00:11:20.516
For the have-nots, c- no payment on outputs and input,

63
00:11:21.476 --> 00:11:39.076
significant loss of traffic from Google, no ability to have a control of my brand and my content's appearance and use in gen AI answers, I really, really worry about the 99.5% of media.

64
00:11:39.956 --> 00:11:52.516
I don't... I worry less about the large players, the, the, the, the nationwide newspaper or magazine who in many instances, particularly in the English-speaking world, have been able to make a deal.

65
00:11:53.256 --> 00:12:04.276
I also worry about those deals going forward. Are they sustainable? Are they just a one-off payment, and then in three years' time people change their mind and say, "Actually, no, we've done this now.

66
00:12:04.616 --> 00:12:17.388
You've got our tech."Because as part of the, the deal, you got our tech. Now it's up to you to use our tech to build engagement with AI, engagement within your own media channels but there is no more deal. Who knows?

67
00:12:17.728 --> 00:12:22.368
I, I'm not a fortune teller, but I really worry about the next--

68
00:12:23.388 --> 00:12:36.128
ab- where we're gonna be in five years' time if the industry that doesn't have a deal today doesn't come together and collaborate to build an alternative, which what I would call a new economic model.

69
00:12:36.308 --> 00:12:49.988
And initially when we started Pro Rata, it was about this new economic model where we build the tech to enable a user to see independent, to have independent attribution on all gen AI answers.

70
00:12:50.128 --> 00:13:05.368
If you can count it, you can share it. Good for the users, we would argue, but also good for content creators, publishers, media owners, because there is a real visible contribution to my content used in a gen AI answer.

71
00:13:05.388 --> 00:13:18.688
But that requires the LLMs to collaborate with an independent attribution company like ourselves. That is not happening today. We've built the independent attribution platform to do that, but- Yeah...

72
00:13:18.708 --> 00:13:29.808
but also to do more. This isn't a tech problem is what I think is my [chuckles] sort of, like, what I would say. It's like it's not a tech problem to understand, you know, attribution. It, it is-- it's possible, right?

73
00:13:29.888 --> 00:13:38.548
Like, I mean, the, the, the, the, the things that AI is being promised to be able to do, I mean, like, every day, and it does sometimes line up with funding.

74
00:13:39.068 --> 00:13:47.288
You know, we, we hear about, like, oh, 95% of doctors will be gone in three years, and, and all kinds of different things that, that AI's able to do.

75
00:13:47.328 --> 00:13:54.048
But somehow, someway, it's not able to actually properly attribute where this stuff came from that they took.

76
00:13:54.148 --> 00:14:06.908
And it's galling, I think, in some ways if anyone has been in media or, or appreciates what media does for society and the economy, and particularly if their jobs rely on it, to be fair.

77
00:14:07.508 --> 00:14:16.627
But at the same time, we have to sort of get on with it. I mean, when I talk with someone like Neil Vogel, I had on a couple weeks ago, you know, he's like, "Look, deal with the devil, I don't know.

78
00:14:17.028 --> 00:14:26.808
We're going in with, you know, it's, it's, it's, it's trust but verify, and this isn't forever. We're not getting married." And, and what I hear is always, "Better to have a seat at the table."

79
00:14:26.828 --> 00:14:36.387
That, that, that is the sort of the, the, the, I don't wanna say cliched answer, but it is kind of the cliched answer. A seat at the table. I've heard this, like, several different times. Yeah.

80
00:14:36.428 --> 00:14:45.058
And, you know, you would argue you would do the same thing. The problem here is we don't have a seat at the table for all content creators. Right. We only have it for what I call the happy view.

81
00:14:45.868 --> 00:14:53.748
Well, the tech people love that. They love the divide and conquer appro- and, and I'm, like, going anti-tech when I... I was just watching Mountain Head, so, like, it's in my head.

82
00:14:53.788 --> 00:15:00.668
But, like, you know, they love to di- that's why a lot of, like, to me, like, technology companies love the creator economy, and they're like, "Oh yeah, why do I wanna...

83
00:15:00.868 --> 00:15:12.118
Like, we spent a generation getting beat up by big publishers. Why don't we just have, like, millions and millions of small people that are dependent on us?" [chuckles] Like, sounds like a better system to me. Yeah.

84
00:15:12.128 --> 00:15:18.388
But I think that's, that's really the risk. But I think you're saying is, like, yes- Mm-hmm... the, the Dotdash Meredith, they're gonna cut their deals, you know.

85
00:15:18.808 --> 00:15:31.648
The New York Times can incur tens of millions of dollars in legal fees, but then, you know, I think some maybe naive people, like, uh, I, I got a couple text messages like, "Wow, a real shocker, they cut a deal on AI with Amazon."

86
00:15:31.708 --> 00:15:38.088
I'm like, "What are you talking about?" I'm like, "Litigation is a form of negotiation," like, right? Yeah, absolutely.

87
00:15:38.168 --> 00:15:50.708
I would argue though that if as part of the settlements they would require i-independent attribution, that the industry overall would benefit from it, and that I'm definitely campaigning for that message to be included.

88
00:15:51.308 --> 00:16:00.708
We're not there yet, but- Yeah... I think it's really important for everybody. You said have a seat at the table. I think it's really important for everybody to have a seat at the table, and there is...

89
00:16:00.738 --> 00:16:08.628
there are comparables in the industry. At S- at Spotify, everybody has a seat at the table. At YouTube, everybody has a seat at the table.

90
00:16:08.748 --> 00:16:20.268
So it's only in gen AI, in gen AI answers, gen AI engagement, that not everybody has a seat at the table when it comes to payouts. Yeah. Do you know what? Do you know something that's endlessly fascinating to me?

91
00:16:20.748 --> 00:16:27.967
Try to get music lyrics on ChatGPT. Go and try to get them. Just any sort of, you know, and you won't be able to get them.

92
00:16:28.748 --> 00:16:35.448
And the reason is the, the-- they've been through this, and they're lawyered up, and they're organized.

93
00:16:35.688 --> 00:16:45.668
Like, unlike publishers who could not organize a two-car funeral parade, which I guess it's not a pers- parade, it's a procession. Maybe it should be a parade, depends on the person. You cannot get...

94
00:16:45.868 --> 00:16:56.168
I tried to get, like, a Tupac lyric, uh, don't ask, [chuckles] and, and I could not get it there. I-- and then, and just, like, randomly, I tried, like I s- I got the lyric...

95
00:16:56.368 --> 00:17:01.458
Uh, what's interesting is you get these lyrics on Google, right? But like I, I took...

96
00:17:01.688 --> 00:17:07.728
I, I couldn't, like, copy and paste something, so I took a photo and I said, "Just give me the, give me the text in here, just the photo."

97
00:17:07.808 --> 00:17:20.928
And it identified that these were lyrics and said, "I cannot perform that function." Wow. [chuckles] Well, well, well. I feel like Columbia now. Yeah. Well, it-- I mean, I think Spotify is a good example.

98
00:17:20.968 --> 00:17:31.778
As the music industry is a good example, and one could argue Spotify brought that product through which the industry had to collaborate. If they had to organize themselves without Spotify- Oh, yeah...

99
00:17:31.778 --> 00:17:44.618
to collaborate on making, making the m- the mu- digital music pay for all players involved and, and, and license it, they would never have been able to do so. Yeah. But Spotify was that independent facilitator. Yeah.

100
00:17:44.648 --> 00:17:53.248
You know, people in media, we've been in this business a little time, people in media like to blame, like, antitrust and all these things, and no, it's just disorganization. They've never...

101
00:17:53.278 --> 00:18:06.038
Any time anything has been to actually have a joint action of any kind, it completely falls apart. It, it, it, like, it's the opposite of OPEC. It's the [chuckles] it's the worst organization. Yeah.

102
00:18:06.048 --> 00:18:17.428
And I just don't see that anytime anyone says they're gonna organize, whether it was Yahoo with their newspaper alliance, you know, so, like, a generation ago, it never, ever works.

103
00:18:17.548 --> 00:18:24.308
And, and so I think it is gonna have to be organized probably to some degree from the outside.

104
00:18:24.628 --> 00:18:33.840
But let's talk about, like, we'll talk about the mechanics there, but, you know, I asked you for a worst-case scenario, right? So give me what a new... economic accord looks like.

105
00:18:33.920 --> 00:18:40.200
I've been like sort of jokingly call it like a Mar-a-Lago accord. Like, what is a Mar-a-Lago accord of...

106
00:18:40.480 --> 00:18:59.220
Which is the, the name given to this idea that there's some grand plan with everything we're doing on tariffs and whatnot to, to create a, a deal that will increase US exports and devalue the dollar and, and increase other currencies and various other things to, to normalize trade imbalances.

107
00:18:59.260 --> 00:19:08.680
What, what would be a sort of Mar-a-Lago accord that resets the economic incentives that are being ripped up right now?

108
00:19:08.740 --> 00:19:24.800
Because I, I don't think people fully organ- realize that Google adopting AI mode is walking away from an economic bargain that has dictated the internet from basically from the commercial internet and, you know, after the portal era.

109
00:19:24.920 --> 00:19:31.140
You know, Google has been the arbiter of the internet. It is distributing, it distributed the traffic.

110
00:19:31.180 --> 00:19:40.760
The unspoken agreement was always that you're gonna get tr- you're gonna allow us to crawl your site, we're gonna take the snippets. It's fine, it's not copyright violation, just pretend it isn't.

111
00:19:40.820 --> 00:19:47.300
And then we're going to send you back traffic that you're gonna monetize. And oh, by the way, it'd be really good if you use our ad tech stack, because we'd like a taste.

112
00:19:47.980 --> 00:19:56.940
Would be terrible for something to happen to that traffic. [laughs] Maybe that's a cynical take, but that's my take. What is a new... What's a Mar-a-Lago accord for the AI era?

113
00:19:58.120 --> 00:19:59.760
So I think it starts with a couple of things.

114
00:19:59.840 --> 00:20:14.560
First of all, acceptance and understanding in the industry that attri- that you can count it, that attribution is needed, that on every gen AI answer, it is possible to attribute the original content sources. And- Yeah.

115
00:20:14.620 --> 00:20:23.960
But they're doing... Right now, like, they are kinda doing it. I even see the rebooting is, is as, uh, as sometimes gets this little logo like up there. Yes. So I'm getting slurped up. I, I don't know.

116
00:20:24.040 --> 00:20:34.560
I was like, oh, I was like kind of excited by that, but I'm not sure if I should have been. But, you know, you know, there is a trust that, that we need to bring into this space when it comes to attribution. Yes.

117
00:20:34.620 --> 00:20:45.480
And that's why we're really working on third-party validation to make sure that our in- our attribution model is independent. I think if you, you know, I, I, I like the com...

118
00:20:46.160 --> 00:20:55.470
You know, going back in time in terms of Google and how it changed the industry forever, and it was a, I think it was a silent killer. None of us really knew- Yeah...

119
00:20:55.500 --> 00:21:03.100
that it was going to be a silent killer, and that's why it was, that's why we can frame it as a silent killer. This time around, it's going much faster, and so therefore...

120
00:21:03.280 --> 00:21:06.340
And, and it's also not very silent, although for so many industry, it is.

121
00:21:07.040 --> 00:21:23.040
Uh, the, so the new accord is educate and inspire all content creators, that educating that attribution is possible, inspire them to, not to be dependent, because we've all become codependent.

122
00:21:23.080 --> 00:21:31.440
You know, and I was a pub- a publisher myself. You become codependent on traffic that needs to come from a third-party source where you have no control.

123
00:21:31.760 --> 00:21:42.200
And I'm really worried about the, the, the Google environment moving into AI, where it is no longer going to be an extension of your brand and your content.

124
00:21:42.260 --> 00:21:53.840
You know, in, in the good old search world, at least your content, your brand was visible, and a snippet of your content as well, so that at least there was some sort of brand and content extension.

125
00:21:53.880 --> 00:21:56.600
That's how you could sell it- Mm-hmm... and then it brings back traffic.

126
00:21:56.680 --> 00:22:17.880
In the new world, there is no brand extension or content extension that you can, I would arguably, I, I would argue, could actually build in, into your business model, into your P&L, because it's out there, there's no traffic, zero click, and search environment brings very little traffic if a, if at all, and there's no forecastable revenue stream.

127
00:22:18.660 --> 00:22:31.759
So what do we need to do to change it? And the industry, what I do hear more and more is we need to work together. And I'm like, well, you've kind of already worked together, but under the control of somebody else.

128
00:22:32.020 --> 00:22:46.340
When a user's looking for an article about the earthquake in Japan, they will get an, they'll get an answer from Google with 10, 20 brands covering that new story. Mm-hmm. In the new paradigm, that's not going to happen.

129
00:22:46.600 --> 00:22:52.000
So but you were effectively working together, you were displayed amongst your competitors.

130
00:22:52.480 --> 00:23:08.860
So our view is build a toolkit that enables you to embrace AI search, embrace AI engagement, not to walk away from it, not to let it, not to leave that to a third party where you don't control. You have no control.

131
00:23:08.900 --> 00:23:13.720
They may change the rules. They may change the algorithm. But actually take back control.

132
00:23:13.740 --> 00:23:31.380
And how do you take back control and collaborate is, you know, and obviously I'm, I'm biased, but I would say that's what we're building in Parata found- with the foundation of attribution, building a toolkit that enables the industry to work together to bring AI engagement within your own channels.

133
00:23:31.620 --> 00:23:41.560
Right. But like attribution is a first step to pay per use compensation. Yeah. Right? Absolutely. So basically, every time, you know, in this like new accord, right?

134
00:23:42.020 --> 00:23:57.310
Anytime, you know, the rebooting is used in a gen AI answer, I get some duckets my way. Is that it? [laughs] Yeah. So a- a- and if you allow me, I can explain what, what I think- Sure...

135
00:23:57.340 --> 00:24:12.100
that the, the, the healthy e- ecosystem is going to look like in five years' time- Sure... if the industry adopts AI engagement tools like the ones that we are developing. So we use licensed content only.

136
00:24:12.140 --> 00:24:25.180
We have licensing deals, and I would argue we're probably only player that a- actively and, and actually licenses content. We use that content to create an AI, ethical AI answer engine.

137
00:24:25.780 --> 00:24:30.760
Some people call it engagement engine, some people call it search engine- Sure... but we could ethical AI search.

138
00:24:31.160 --> 00:24:50.016
An ethical AI search always comes to the independent attribution, always with the link back to the original content source, and it comes with monetization, or part monetization for a moment, but we, we believe that we can only do this, you can only create a healthy ecosystem-If you bring new monetization into the mix.

139
00:24:50.676 --> 00:25:00.096
The toolkit enables us to offer this to publishers with whom we have licensing deals to start to build AI engagement within your own channels.

140
00:25:00.396 --> 00:25:12.806
So you effectively get a toolkit to build ask me anything related links within your pages to drive AI engagement. Because as I, as I said earlier, AI engagement is not gonna go away.

141
00:25:12.856 --> 00:25:26.456
That train's left the station, a billion users across the globe, so publishers, content creators need to embrace it, and we help them to build that AI engagement. Some people have already done that.

142
00:25:26.836 --> 00:25:39.016
Most people who do it have used their deals, the zero point five percent happy few who've, who have received GPU credits as part of their deals. The ninety-nine point five percent

143
00:25:40.096 --> 00:25:55.616
some-sometimes do not know how to do this, can't afford to do this, and, and in many instances, um, do not have enough content if they were even able to do it, do not have content to deliver desired answers by users.

144
00:25:56.116 --> 00:26:02.186
So what we came up with is this toolkit that you embed within your pages, drives AI engagement.

145
00:26:02.556 --> 00:26:18.456
It uses content from not just your own corpus, but also the corpuses that we have licensed with our trusted content partners. Okay. And it should say, I mean, ProRata is started by Bill Gross, who has a long history.

146
00:26:18.576 --> 00:26:22.476
I, I... everyone should know who Bill Gross is, at least in my view. I think so too.

147
00:26:22.756 --> 00:26:32.436
Because, you know, he's really the person who pioneered the greatest economic engine that the world has ever seen, which is known as paid search. Google did not invent it.

148
00:26:32.476 --> 00:26:46.256
They actually had to pay for ripping it off [laughs] if you really go back. Because, you know, what was Overture at the time, which actually was rebranded. I, I, I go all the way back, Annelise. I go back to GoTo.

149
00:26:46.356 --> 00:26:58.116
Me too. That's how you know the OGs. They go, they go back to GoTo, not, not Overture, you know? [laughs] And, you know, I think at the time, I... it was really when I started, it was really when I started in journalism.

150
00:26:58.436 --> 00:27:07.256
I remember having a meeting with... and it wasn't with GoTo, it was with FindWhat, which was one of the other competitors. I know I'm going way back, right?

151
00:27:07.276 --> 00:27:17.356
[laughs] And I remember them explaining this to me, that they were putting these ads on search results pages. I was like, "This is insane." I'm like, "You can't do this." That was, like, what it was back then.

152
00:27:17.696 --> 00:27:24.136
Nobody understood that the search results page was the most valuable piece of real estate in the world.

153
00:27:24.596 --> 00:27:34.896
It was used for remnant banner ads, like, whatever, the four, four eighty-six by sixty, whatever the, the, the little banner ad size was back then. Maybe some skyscrapers.

154
00:27:35.716 --> 00:27:42.096
But it-- nobody understood it until GoTo, and then Overture started the, really the paid search system.

155
00:27:42.176 --> 00:27:52.236
Now, I mean, you could argue Google really perfected it and really, you know, was able to use its muscle in order to, you know, basically take over the entire market.

156
00:27:52.516 --> 00:28:02.946
And it's, it's interesting to me, like, because of all of the, the antitrust actions now, I'm, like, reading through them, I'm like, "Yeah, they've been doing this, like, since the beginning.

157
00:28:03.056 --> 00:28:13.626
This is, like, the business model. What are you talking about?" I was like, "I'm talking to Overture, like, every other week because Google was cutting uneconomic deals with AOL to buy distribution."

158
00:28:13.656 --> 00:28:20.156
It just did it with Apple. Like, it... they literally have been doing this since the start, and Overture couldn't compete because the...

159
00:28:20.336 --> 00:28:29.776
that was their business, and basically, Google was saying, "We're gonna give you more money than you generate because we're gonna, we're gonna wipe out the entire market because we have our own search engine."

160
00:28:29.836 --> 00:28:40.736
All that is a long preamble to say w-what I'm sure that that experience wa-wa... it was top of mind in designing ProRata for a new, a new era.

161
00:28:41.116 --> 00:28:52.256
But explain how it was built in such a way with, with that experience in mind. Little history, [laughs] little history- Yeah... lesson there for everyone. [laughs] Yeah.

162
00:28:52.416 --> 00:28:58.376
You know, it's, it's lessons from history and the things that we don't-- did we want to repeat and we, we don't ever want to do again.

163
00:28:58.866 --> 00:29:08.616
And I think i-in the show notes, definitely if you can put a link to Bill's Idea Lab where everything on the website- Oh, yeah... you need to know about Bill is there.

164
00:29:09.156 --> 00:29:20.076
You know, he's a legend, and he's a legend in many ways. He has a, a real talent for understanding where there's a gap in the market and where there's an opportunity to n- to build a new business model.

165
00:29:20.596 --> 00:29:29.086
So when he read the New York Times, when he read the news about the New York Times lawsuit, he thought, "This is just not right." You know? In, in...

166
00:29:29.146 --> 00:29:41.156
within Gen AI answers, only a few people are, are able to take the large language models to court, and only a few will be able to do a deal. But what's out there for the rest?

167
00:29:41.376 --> 00:29:47.996
I need to build a new economic model where basically if you can count it, you can share it. That's how this all started.

168
00:29:48.076 --> 00:30:02.456
And by setting out the mission that I want to build a business that is pro-publisher, pro-content creator, pro-author, and I want to help them and guide them to speak up for themselves and, and, and to highlight where their content is used.

169
00:30:02.516 --> 00:30:12.396
We build a, we build a, a team around this mission that actually comes with deep tech ind- experience, deep media experience, and deep entrepreneurial experience.

170
00:30:12.436 --> 00:30:23.826
And on the entrepreneurial side, I mean, our tech team built the Bing knowledge graph, the Bing advertising system. Bill built GoTo, Overture, which basically now is, what is it?

171
00:30:23.976 --> 00:30:36.446
Three hundred billion dollars of revenue across the whole, and probably even more. So he's, he's, he's the founder of these new implementations of what I would call old ideas and, you know- Mm-hmm...

172
00:30:36.516 --> 00:30:39.656
a, a traditional existing ideas. You know? If you build...

173
00:30:39.816 --> 00:30:47.816
if you make something great, you'll, you'll get an audience, and then you can get paid for it, whether it's the audience pays for it or the advertiser pays for it. Yeah.

174
00:30:47.836 --> 00:31:04.669
And in, in this new world of AI, we need to be able to replicate it, and for that, we need to understand the tech, and we need to build the tech that enables us to build great product-You know, and I think Neil said it so well in that podcast, you know, it's all about building great- Mm-hmm...

175
00:31:04.720 --> 00:31:10.540
and engaging products. I loved it when he said, "A media business is fantastic. You build something great.

176
00:31:10.580 --> 00:31:20.070
You build a great content story, great storyline, a great product to engage with that story and that content, and then you get an audience. And then you monetize the audience."

177
00:31:20.640 --> 00:31:29.140
And we can do the same thing in the world of AI, but we need content owners and creators to understand how to do this, and that's where...

178
00:31:29.690 --> 00:31:43.040
That's, that's the f- the foundation of our company, which is an economic model, a very experienced team who's incredibly passionate about solving this problem, and we do that with the industry.

179
00:31:43.260 --> 00:31:54.060
You know, we lean in- Yeah... we, and, and, and that's... Together we can solve it, and I always say to publishers, "I don't have all your answers, but I have the best team to get to the right answer." Yeah.

180
00:31:54.460 --> 00:32:01.120
But I guess the point I was trying to get at was, like, are, are you building a consumer company or a B2B company, or both?

181
00:32:01.520 --> 00:32:09.920
In that, like, I get the, the positioning of, like, we're on the side of publishers, you know, and, uh, it's why I wanted to talk to you [chuckles] to be honest.

182
00:32:09.930 --> 00:32:17.160
So if you were like, "I wanna put them out of business," I'd probably be like, "Yeah, let's pass on Elise." Yeah. Yeah, yeah. Um [chuckles] you know, I, I like that, right?

183
00:32:17.240 --> 00:32:26.690
But I go back to the Overture experience, and Overture, you know, look, it got bought by Yahoo eventually. It was a, it was a great exit. But it did, it lost against Google, right?

184
00:32:26.720 --> 00:32:34.820
Because Google had a search engine, had a direct tie to the consumer, it vacuumed up a ton of data, and it overwhelmed Overture in the marketplace.

185
00:32:34.940 --> 00:32:44.980
It just slowly lost every single deal because Google could say, "Name the number, we'll just take it." I mean, the, at the time, people don't... And, and Google was tiny compared to how it is now.

186
00:32:45.100 --> 00:33:00.840
I can remember, go back to Neil, About.com at the time, which is the sort of origins of Dotdash Meredith, you know, they used a rival AdSense product called Sprinks, I believe, and About was one of the largest sites on the internet.

187
00:33:00.980 --> 00:33:09.050
Google was like, "We need that real estate." So they just bought Sprinks and shut it down. [chuckles] It's like, here you go. [chuckles] You're using AdSense now.

188
00:33:09.760 --> 00:33:23.060
You know, and that, I think, spoke to, to how things were going. But like, I, I'd love to get at that, because, like, explain to me, is Gist AI gonna try to compete with Chat- with ChatGPT and, like, and, and Gemini?

189
00:33:23.260 --> 00:33:33.180
Like, that seems like a lot. No, we built Gist AI because we needed to prove that attribution's possible- Okay... and we needed to show to publishers what it looks like. What is...

190
00:33:33.240 --> 00:33:44.980
You know, we're using your content to build an AI answer experience, and this is what it looks like. That was step number one, and now we're working with these publishers to actually take Gist

191
00:33:45.880 --> 00:33:59.599
into a, we call it Gist-powered answer, GPA product, that you integrate across your channels to drive distribution. Okay. So it's mostly... So explain how, how do you get leverage in that? Because I think the, the...

192
00:33:59.640 --> 00:34:05.920
You know, we can talk about what's fair and what's not, and, and it's, it's interesting, and I'm, I'm probably gonna e- end up agreeing with you that it's unfair.

193
00:34:06.560 --> 00:34:19.160
But at the same time, I don't know, my experience in, in the world [chuckles] has generally been, it really comes down to leverage. And publishers individually have, like, no juice. They have no leverage.

194
00:34:19.360 --> 00:34:29.740
I mean, good luck. You might as well be, you know, going up against the, the house of Lannister with, like, a rock. It's just not gonna happen. Or, or not even the Lannisters, the ones with the dragons.

195
00:34:29.820 --> 00:34:38.180
So, like, that, that's out. And then I don't even know, first of all, the history of publishers doing anything together is very consistent in that it never works.

196
00:34:38.780 --> 00:34:52.300
And so I end up wondering, why in the world would, outside of some kinda legal action, which does not seem, which is not happening in the United States, obviously, at least in the next three-plus years.

197
00:34:52.820 --> 00:35:01.879
You, you look at the, the bill going through, like, right now, I mean, it, they're protecting the AI companies from this. And so that's not gonna happen. No one's coming, coming to the rescue.

198
00:35:02.260 --> 00:35:15.400
But then it leaves me to wonder, it's like, what is the pa- what is the leverage that publishers can have other than saying please, and then pretty please? Okay. The leverage is, it's the power of the numbers.

199
00:35:15.640 --> 00:35:23.960
We've done almost 100 deals, uh, agreements I call them, they're not deals, agreements. We signed agreements with a- almost 100 publishers.

200
00:35:24.020 --> 00:35:32.520
And we have many non-English publishers kind of, like, ready to sign, but we're, we've, we're focusing on English language for the first half of this year.

201
00:35:33.100 --> 00:36:00.380
So the leverage is the power of the numbers, the collaboration across the industry, and every publisher that signs up today with us already signs up to collaboration, because my content is used alongside content from my competitors, adjacent neighbors, and so it is already a collective product through working with a large number of publishers and many, many more to come.

202
00:36:01.180 --> 00:36:04.530
So the leverage is power in size. If I look at

203
00:36:05.480 --> 00:36:36.760
the ex- the, the extent of the network, you know, you can count it through page views, number of titles, but where Gist can be Gist-powered answers, the AI engagement powered by, by our product can take place is across a network of all our partners, because our partners sign up to license content, but effectively they also sign up to get a suite of AI engagement tools that they can implement within their channel, within their own media channels.

204
00:36:36.770 --> 00:36:47.420
So I think the leverage is not gonna come overnight, but there's an incredible signs across the industry. There's an appetite to collaborate, but you can't collaborate through negotiation.

205
00:36:47.520 --> 00:36:59.420
So what we bring is collaboration without negotiation through a product solution. Okay. And how does this change how publishers make money? Are they still gonna... 'Cause I think that's what I end up wondering about.

206
00:36:59.460 --> 00:37:06.920
Like, the-Like, there's so many problems that are created by this new paradigm of people getting answers, right?

207
00:37:07.300 --> 00:37:14.800
But one of the things that seems to me, that's why I'm, I'm always asking people are people gonna be going to your website in five years?

208
00:37:15.560 --> 00:37:20.199
And I don't mean it to be a jerk or something, but I think every, every single person needs to be thinking that.

209
00:37:20.240 --> 00:37:35.050
Like, honestly, with the changes going on right now, are people seriously going to be going to www.yoursite.com and then hitting a back button when they're done? Like, it just seems strange to me, and I just wonder...

210
00:37:35.680 --> 00:37:44.300
A-and then you're gonna be monetizing it through programmatic ads? I mean, really? Like, that just seems like a very fraught proposition.

211
00:37:44.340 --> 00:38:04.840
So there's so many issues to, like, solve, but, you know, one thing that I always wonder is whether a lot of publishers will end up being wholesalers in that they're wholesaling information to the aggregators which are going to be these answer engines and whatnot, that they might not even have a quote, unquote website.

212
00:38:04.900 --> 00:38:09.180
A website is just a front end to like a data depository at the end of the day.

213
00:38:09.220 --> 00:38:28.500
Like, shouldn't-- isn't there, like, a case to be made that if the economic bargain can be made, a lot of people will be more in the licensing business than they will be in this, you know, this business where it, it doesn't re- it's so convoluted in tr- how a lot of publishers have to make money.

214
00:38:28.780 --> 00:38:41.920
Like, these businesses are overly complex for their size, and they're getting smaller, and when you have complexity on top of shrinking, it's just a mess, and it's because the ad system is broken.

215
00:38:41.960 --> 00:38:54.600
It has been broken almost from the start, and subscriptions are only, like, gonna do so much. Commerce was an SEO strategy. Forget that. Get ready for every com-- not every, most commerce business lines.

216
00:38:54.640 --> 00:39:04.040
You're not gonna be hearing a lot of them [chuckles] going forward, 'cause people are gonna be quietly taking them upstate. I mean, BI just took theirs upstate, and, and others will follow, I'm sure.

217
00:39:04.080 --> 00:39:16.340
And so what I wonder is, is long-winded set way of saying, do you see like a possibility where how publishers make money will increasingly be getting, you know...

218
00:39:16.610 --> 00:39:33.700
It's, it's more licensing than it is retail, and by that I mean, like, getting people to come to your website, having, you know, figuring out all the programmatic and, like, all of these different things that, that, that are being done for businesses that frankly are shrinking.

219
00:39:34.820 --> 00:39:42.020
Wow. So how do I, how do I save the industry in five minutes that you raise- Be concise, Annelise. Yes.

220
00:39:42.310 --> 00:39:58.700
So licensing needs a fair economic model, and I think if you look at the media industry as a wholesaler of content, if you license that without brand, without mentioning and contribution of, of brand and actual content itself,

221
00:39:59.640 --> 00:40:14.370
y-you know, I don't want to go back to the time, the time when I lived in, in Russia and I was a magazine publisher, and we had about twenty magazines, and that, that was a, an innovative landscape because there, under Soviet times, there were only two magazines.

222
00:40:15.020 --> 00:40:20.640
You know, consumers actually like- Yeah... new media, and they like media that comes with a brand.

223
00:40:21.140 --> 00:40:32.570
And if we license it and it's all vanilla, and it comes out with brand and any contribution to an, to your original source, I don't think that's a very exciting media- Yeah... landscape to live in.

224
00:40:32.580 --> 00:40:39.660
It's not the one that Neil was d- was describing in your podcast two weeks ago. So but yeah, is, are people still gonna come to your website?

225
00:40:39.760 --> 00:40:47.260
Some of them will do it, but, you know, we've seen the website traffic is- Yeah... decreasing significantly. Some people get magazines too. But yeah. But people still gonna...

226
00:40:47.340 --> 00:40:57.170
So it is as a publisher, you need to lean into product development, and that product development goes across a number of channels or a number of ways to reach your- Yeah...

227
00:40:57.280 --> 00:41:07.090
your, your audience, and that audience is now used to having an AI experience as well, at least many of them are. And for some of that, that may be an article summarization.

228
00:41:07.120 --> 00:41:25.260
For some of that, it may be a button where I can ask a question because I happen to have a question, and then the publisher can decide for that question to be answered by just my content or actually content of a number of trusted content sources because I know I can do a better job for the consumer.

229
00:41:25.300 --> 00:41:35.060
Yeah. So I think it starts with the consumer, and the consumer has already shown to us that they can navigate different channels. They can go from a podcast, and now we have podcasts on YouTube.

230
00:41:35.400 --> 00:41:48.400
So we need to really follow that consumer u- consumer desire and behavior and build for it. And I'm not a publisher today, but I deal with them all the time, and I like to inspire them wherever I can is to...

231
00:41:49.040 --> 00:42:00.280
AI engagement is another user experience that you need to build within your workflow, and it's there to drive more engagement and more engagement to then be monetized. Right.

232
00:42:00.480 --> 00:42:06.799
But people have to come to your website in order for them to ha- engage with any of these things. Like it's- Well, not, not if you...

233
00:42:06.860 --> 00:42:27.860
I mean, what we're building with kind of the AI toolkit just powered answer is imagine a user goes to The Guardian's website, uh, The Guardian's app in London and wants to know about y- their upcoming Yellowstone holiday, and they ask a question like, "What do I take on my holiday to Yellowstone, and what are the top things I should avoid when I'm there?"

234
00:42:28.440 --> 00:42:40.180
That answer could just you look at The Guardian's corpus and, you know, you're lucky and a, a journalist wrote an article three years ago about a travel, their travel story to Yellowstone.

235
00:42:40.220 --> 00:42:50.200
If you combine that corpus with the corpus of the Billings Gazette, which is practically next to Yellowstone, you get an answer that's based on recency.

236
00:42:50.240 --> 00:43:12.968
So you get depth and breadth within the answer, and it doesn't matter for the Billings Gazette that that content is used for a user in London.Who reads The Guardian or a user in the UK who reads The Guardian wants to know about Yellowstone because through attribution there will be compensation for users of that content in a completely different media channel.

237
00:43:13.588 --> 00:43:21.148
Yeah. So you've, you've worked... I guess they used to call it Fang, now we're on the Mag Seven. But you worked at, you worked at Facebook. I did. Like, so you have experience in these companies.

238
00:43:21.188 --> 00:43:30.748
And I don't mean to generalize, but I will generalize. That's, you know someone's gonna generalize when they say that. Talk to me about your experience about how those companies in general, and 'ca-

239
00:43:31.778 --> 00:43:42.048
ca- and I've always felt this divide between Silicon Valley and New York in my career. And, and again, I'm just, like, speaking in very broad generalities. But it's basically between, like, tech and media.

240
00:43:42.088 --> 00:43:57.128
It's like, you know, men are from Mars, women are from Venus. It's, like, really kinda is that kinda thing. And, you know, you're, you're a media person, right, too? [laughs] So what, what sort of do you think that...

241
00:43:57.708 --> 00:44:12.968
I don't know. I'd love to hear your takeaways about how broadly speaking the world of tech sees publishing and media, because I think that really matters in how they approach these, these problems.

242
00:44:13.008 --> 00:44:18.288
I mean, the- they're not, quote-unquote, media people. A- and that's, you know, I, that's what I said.

243
00:44:18.328 --> 00:44:24.968
I was, you know, I was, I was watching that, that Mountain Head thing, you know, which I didn't really like personally that much. I didn't think it was that funny.

244
00:44:25.048 --> 00:44:37.548
But, you know, it, it does sort of drive home that, that point that, yeah, these people think a little differently. Yeah. First of all, I'm, I'm incredibly... I was incredibly excited to join.

245
00:44:37.588 --> 00:44:45.208
I was not joining on the media partnership side- Yeah... but I joined on commerce because I've built many classified websites in my career.

246
00:44:45.328 --> 00:44:58.568
I think you only understand what product-led companies, what that means once you're there and once you're in the midst of it. It, a company that's purely focused on building product for engagement and traffic.

247
00:44:59.468 --> 00:45:06.328
And when you do that at such a large scale, there are some things you, there are some elements that are really crucial to us.

248
00:45:06.388 --> 00:45:14.648
You know, if we're a media person, like, you know, the brand of The New York Times or the brand of The Washington, Washington Post, we, you know, we can distinguish between these two different brands.

249
00:45:15.208 --> 00:45:26.368
But when you work on product at, and with, with such, for such a large audience, and you're an engineer, you probably have a different... You know, it's the alphas and the betas.

250
00:45:26.428 --> 00:45:35.968
You, as, at least that's how we classify people in the Netherlands. You're, you're an alpha person, you're a language person, you're a beta person, you are a, a, a STEM person.

251
00:45:36.748 --> 00:45:44.538
And there's a different appreciation- Wait, who are the alphas? Are the alphas are the, are the STEM people? No, the alphas are the... No. So that's why- Oh, really?

252
00:45:45.108 --> 00:45:55.848
When I say, when I say alpha, beta, it, it is definitely a direct translation from Dutch. [laughs] So the alpha people are the people who are the linguists. Oh, great. And- I'm an alpha in Du- in, in Netherlands...

253
00:45:55.868 --> 00:46:06.828
and, and, yes, you are. [laughs] And the beta people are the STEM people. Good. So there is a, there is a- I like this. I'm moving. And actually I have... No, I actually have two STEMs at home.

254
00:46:06.888 --> 00:46:09.127
My two kids are both the STEM kids.

255
00:46:09.148 --> 00:46:22.308
And, and actually that was a, a great, uh- Yeah, you gotta de-risk, you know, you gotta [laughs] That was, that was very helpful in my career because you need to look into the brain of s- somebody who's an engineer.

256
00:46:22.808 --> 00:46:29.358
Yeah. And what is important for them, it's code, it's numbers. Mm. It's not words. Or words matter less. Mm.

257
00:46:29.408 --> 00:46:43.788
So, and I always say this to publishers, if you negotiate with somebody who has a different perception and different sense of respect for what you're making, I'm not sure if you can negotiate through a sense of...

258
00:46:43.948 --> 00:46:51.108
You're, you're not equal and you're not, you're not pursuing the same- Yeah... outcome, or it's not based on the same principles.

259
00:46:51.648 --> 00:46:59.368
And so my, what I loved about working in Silicon Valley, and actually later on I moved to LA because I, I, I...

260
00:46:59.388 --> 00:47:15.408
And the way I described it is, you know, people in, in Silicon Valley are very excited about new technology because of the tech itself and, you know, the, the products you could build and then you move to LA and it's like, what does that tech do for, for me in terms of money?

261
00:47:15.648 --> 00:47:25.028
Yeah. You know? A hundred percent. It's like what, what's the use case, you know? So a, a, a cool piece of tech, but actually is it useful? And if it's not useful, but it's really cool.

262
00:47:25.808 --> 00:47:33.928
So a commoditization of things that we, that are really precious to publishers, I think, is where the big divide is.

263
00:47:34.588 --> 00:47:47.448
And that's, you know, Justine Roberts, founder of Mumsnet, described on a podcast how she was so insulted during a negotiation with one of the LLMs where they described her content as tonnage. Like def- Tonnage?

264
00:47:48.088 --> 00:47:55.728
Tonnage of your, tonnage of your content. She's like, you know, she was enraged. Yeah. And, you know, I am enraged about it as well because...

265
00:47:56.128 --> 00:48:01.548
And that's, you know, comes back to- But to an engineer, it's just a bunch of data. Yeah. You know, that's why- [laughs] It is, right?

266
00:48:01.668 --> 00:48:12.748
You've heard them say, you know, "We've, we've got enough humanly created content, s- so now we can just use synthetic." Yeah. You know, th- to me it was, like, always like...

267
00:48:12.808 --> 00:48:17.838
And this sort of went away, but for years it was, like, talking about what is premium, right? Like, what is premium?

268
00:48:17.928 --> 00:48:24.108
And to, like, publishers it's like, you know, it's like the old thing they would say about pornography or obscenity. It's like you know when you see it or whatever.

269
00:48:24.148 --> 00:48:32.168
Like, you know, there's no te- but he's like, "You know premium." But in the, and they're like, but in the engineering brain, they're like, "Who are you to say that it's premium?" And like, "Are we talking about prem..."

270
00:48:32.268 --> 00:48:42.568
I remember YouTube started talking about pre- premium content, premium vi- premium audiences. Like, and there's, there's all kinds of ways to twist this, and I can, I can kind of understand from that.

271
00:48:42.668 --> 00:48:55.748
I came from an engineering family, at least, and so I kind of understand that, that mindset and, and particularly being product-focused, where the externalities i- are, are, are not, are not your main concern.

272
00:48:55.808 --> 00:48:58.508
Like, I, I did this, I did, I do this podcast, People Vs.

273
00:48:58.548 --> 00:49:11.708
Algorithms, and one, um, one of my co-hosts, Alex Schleifer, he, he was an executive at Airbnb and he ran design there and he was like, "We're making changes that would affect literally millions of people.

274
00:49:11.788 --> 00:49:15.128
It would affect like, you know, tens of thousands of people's, like, livelihoods and stuff."

275
00:49:15.708 --> 00:49:23.862
And he was like, "We're just, like, moving pixels around."And the reality is like you don't think about that, and not because like you're like heartless or something.

276
00:49:23.902 --> 00:49:32.452
I know Alex is not heartless, but just because like that's the job, you know. And, and that is the approach because if you're gonna be product-centric...

277
00:49:33.372 --> 00:49:40.732
And, and this is, you know, this is in Mountain Head, and this is, you know, just the reality I think of, of a lot of these Silicon Valley companies. When you're operating at scale,

278
00:49:41.872 --> 00:49:51.712
y-you're gonna, you're gonna run over stuff. You know? I mean, imagine like running with these companies when like, you know, Facebook is being accused of, of, of fomenting genocide in Myanmar.

279
00:49:51.772 --> 00:50:04.372
Like, I mean, like these things are like real that they deal with that are far beyond, you know, people's precious webs-webpages when they're building these products at such massive scale. So I don't know. I kind of...

280
00:50:05.052 --> 00:50:12.812
I just think it's interesting to, to think about like, you know, from, from the different perspectives because when you go into these things, it's better...

281
00:50:13.381 --> 00:50:25.172
It's not good to be naive about like what, what the objectives are. And like, I think for too long, too many people in publishing in the last era thought that they could... If they could just...

282
00:50:25.212 --> 00:50:36.752
Like Jonah Peretti, if he could just convince, you know, Mark Zuckerberg about how important, you know, this content is to like Facebook. It's like, come on, Jonah. No. Like, I mean, that's just- No. Yeah.

283
00:50:36.832 --> 00:50:43.632
I th- I think it is almost a definition of insanity because you tr- do the same thing over and over again and you- Yeah... expect a different result.

284
00:50:43.672 --> 00:50:55.222
Like, when is, when is big tech going to appreciate my content and my brand? And but one of the key things I've learned is how to scale. And, you know- Mm...

285
00:50:55.252 --> 00:51:02.712
in, in, in a, in an environment in a company like Facebook, it's in- it's incredible. It's like you're in a sweets shop, and you build a product for...

286
00:51:02.752 --> 00:51:15.472
You know, you pilot it with 25, and then you find the right pipes, and you push that content, get that product through 10 pipes, and, and then it works for 100,000 people, 100,000 merchants in that particular case.

287
00:51:15.812 --> 00:51:23.192
So you build it for 25, and then you have it in the hands of 100,000 merchants in, in a year's time. That is exhilarating. [chuckles] And I've taken that...

288
00:51:23.352 --> 00:51:29.652
But I've taken that experience with me into Pro rata because you asked me about it, what is the leverage?

289
00:51:30.032 --> 00:51:45.372
And I think the leverage we have is if we can bring our product into the hands of hundreds of publishers, maybe even thousands of publishers, the publishers can start to build decentralized AI engagement in their own channels, but through a collective.

290
00:51:45.412 --> 00:51:53.721
Because if, if, if the content I have for an AI experience is not enough, then I have all of Jess content, the content from the trusted network.

291
00:51:54.192 --> 00:52:06.712
And so I, I deeply believe that by, by building product for the industry to work together- Mm-hmm... we will build leverage in the industry, but it needs to come with money. It needs to come with monetization.

292
00:52:06.772 --> 00:52:13.881
It isn't just distribution and, and more engagement throughout the network and more engagement on my own channel.

293
00:52:13.932 --> 00:52:22.962
It needs to come with monetization, and that's why I'm so super excited about the, the ad product we've built, and it's kind of like full circle for Bill as well because- Mm-hmm...

294
00:52:22.962 --> 00:52:24.952
that's why he started to build an ad product.

295
00:52:25.392 --> 00:52:38.852
This ad product goes back to my days when I was a publisher and I was selling Men's Health's next issue, and there was a feature on jeans, and I would definitely call Wrangler and Le- Levi's and Calvin Klein.

296
00:52:39.552 --> 00:52:54.512
And we've moved away from that experience where the ad is really performs best when it's contextual with semantic understanding, and that's what we've built for the ad to... And it's an ad that lives within the answer.

297
00:52:54.572 --> 00:53:07.092
It lives within the AI engagement and has a real understanding of... Think about it as an agent that can read all the answers and can read all the brand information and combines- Yeah... that into a new ad product.

298
00:53:07.372 --> 00:53:20.482
But, you know, I just wanted to, to, to raise that because as we know, collective a-action or people are bringing people together, they need to understand at the end of the li- at the end of the day, how is this gonna pay for my journalists?

299
00:53:20.512 --> 00:53:30.932
Well, yeah. Yeah, I mean, 'cause like I don't think the licensing revenue is gonna be enough, and not for most, right? And i-if they're able to get it. And the ad experience clearly needs to change.

300
00:53:31.132 --> 00:53:44.202
Like, I mean, I don't... I've talked about this repeatedly on this podcast. Like, it's a disaster, so might as well like start all over again. Like, I don't... Like more, you know, pop-ups were a, a, a disaster.

301
00:53:44.202 --> 00:53:55.932
Retargeting was a disaster. Like, everything led... It's kinda funny 'cause in some ways like, you know, tech enabled a horrific user experience that then, you know... Uh, this is a common theme.

302
00:53:56.272 --> 00:53:59.192
Create a problem, and then you solve the problem that you created.

303
00:53:59.432 --> 00:54:15.592
And, you know, it's like, yes, people are going to opt for some kinda answer engine than clicking on a link on Google and going to some horror show of, you know, like, uh, a UK local publisher site will like make your eyes bleed and your computer overheat.

304
00:54:16.292 --> 00:54:23.092
So you know, and I understand why that happens, but the ads have to, you know, change. And you know, I...

305
00:54:23.132 --> 00:54:27.232
Look, Cannes is in a couple weeks when we're recording this, but there's gonna be a lot of people in Cannes who are...

306
00:54:27.302 --> 00:54:36.972
It's gonna be all about AI creative and automating [chuckles] the creative, creativity that this, this festival is supposed to be celebrating, and there's gonna be no...

307
00:54:37.032 --> 00:54:47.392
That is gonna be lost on everyone, uh, as they sip rosé, the fact that this literally is supposed to celebrate human creativity, and it is gonna be all about [laughs] how technology- Yeah. Yeah...

308
00:54:47.422 --> 00:54:59.352
can replace human creativity. I mean, I w- we're running out of time, Brian, but I- I know... am so passionate about this because ads should be great. I mean, advertising is wonderful.

309
00:54:59.452 --> 00:55:09.072
It should be a joyful experience that then inspires you to- Yeah... to, to buy something. And it's y- I think we've totally moved away from it for many reasons you just mentioned.

310
00:55:09.692 --> 00:55:21.392
But we need to bring that back, and what I hear about AI in ads, and there's so many new players, so many new products, it is almost always focused on the optimization of the creation.

311
00:55:22.132 --> 00:55:32.432
And very few times do I hear people ta-talk about the optimization of the experience for the user. Yeah. How do we use AI to make it a better user experience, to make it delightful?

312
00:55:32.492 --> 00:55:44.892
As Bill would say, "Thank you for this ad." That's what we're... You know, that's the spirit of what we're building. We want to build ads that, that make users say thank you. This is really useful. This is the ad I want.

313
00:55:45.402 --> 00:55:56.402
And isn't that where advertising started in the first place? Yeah. Okay. Thank you for this, Annelise. This is a real pleasure to, to go through these very important issues, and I think mostly optimistic.

314
00:55:56.432 --> 00:55:59.432
I think there's some optimism- Yeah... in there. Oh. 'Cause I think that there's...

315
00:55:59.472 --> 00:56:10.532
Yeah, this is a, this is a period where, you know, publishers can't play defense forever, and, you know, it's time to, you know, chart a, a, a more, as well as you can, an independent path.

316
00:56:10.652 --> 00:56:25.912
And like whether there can be collective action that makes clear the leverage that quality content has, you know, that the, the world actually needs it and wants it and that these AI engines need it, you know, hopefully that will, you know, lead to a better outcome.

317
00:56:26.252 --> 00:56:37.712
But thank you so much for taking the time. Thank you, Brian. See you in Cannes. [outro music]