Like when they do the New York City apartment walkthrough, like, you won't believe this 600-square apartment or 400-square apartment has two people and all this cool stuff in it. I've seen that, but I've never actually been in one. Yeah.

Oh, I can probably rent it out for about $1K a month if I wanted. Real estate's crazy these days, but it might be a little unethical.

Eric's like, from my New Jersey beachfront mansion, I scroll on the Instagram, and it shows me how the poors are living in the urban environment. For the

record, I never took the ladder away while he was sleeping. That's good. It's an extension of trust. All right. So we got the insult out so the pod can start.

It's a team building exercise. All right. Go ahead, Joe. Welcome to Open Market. You know what you're listening to. It's a podcast about building companies in the advertising industry. Eric and I are joined by Michael Bishop, the co-founder and CTO of Open Ads. Welcome, Michael.

Thanks so much for having me, guys. I'm excited to be here. You're

like the voice of God. Oh, this gets ironed out in post-production. Otherwise, you're going to sound way better than we do. The audience can't see, but Michael has a very professional sounding mic. Like, you should probably just run the podcast. Michael, tell us what we're talking about today. Just kidding. Joe, what's going on? What are we doing? We want to talk about AI and the future of ads. But Michael, just like we did with CERN a couple weeks ago, fellow AI startup founder in the Admore tech space, why don't we just talk about by diving into the business a little bit? Why don't you tell us about open ads?

Sure. Put it simple. We are creating and delivering unique ads to every impression in real time. Got to dive into that a little bit because I think the first question is, how is that different from dynamic creative optimization? And the way I would put it is DCO is picking from a finite number of variants. And what we're doing is tapping into the ability for AI to make a completely unique ad. So given an impression, a context, a user, that has never been seen before. And that's something that was not technologically possible before large language models, and honestly was not feasible financially or inference time at scale until even the last couple of months even.

Michael, there are a few different ways that generative ads could manifest. It could be on generative AI platforms like OpenAI. It could be like against the content contextually. It could be on publisher content, especially as that content itself potentially becomes more dynamic with AI. Or it could just be, you know, anywhere on the internet. And it could just be, as you said, sort of the next generation of DCO. Like instead of picking from options, you just have a completely generative ads. Which of those opportunities are you pursuing or are you most excited about or seeing traction with? Pursuing

all of the above. And one thing that I should have mentioned previously is not just generating ads, but we are stitching LOMs into the ad creative itself. Like what we have is like actually like branded assets, the creative voice of the brand in their product data, their marketing collateral, and whether it's like suggested questions like or freeform follow-up chat, you being able to just like pull for information from the brand from an ad. In terms of breaking down these sort of platforms this can serve on, there is definitely the like consumer-facing generative AI opportunity, which I think you could describe as a hybrid between search and the open web, where what you have is users putting in effectively a search string, giving you the context for exactly what they're looking for, and the ability to transact on it, not just in a rigid keyword basis, but completely freeform. What is the best ad based on the user intent, what we know about them, what you can extrapolate about this opportunity. And like one of the really cool things we've been able to test is not just advertiser selection, but testing against like regenerated static creative versus tuned in real time to that search input. That real time personalization leads to more than 50% higher click through rates. Like consumer pacing AI is one thing. Traditional publishers and traditional media, you can treat it as an expansion to contextual targeting. We're definitely seeing some interest from some of the major news and media orgs around this as a high-value ad unit that they can upsell to their advertisers as something premium, as something super highly engaging, as very custom branded experiences. So seeing a surprising amount of traction there. And sort of the third pillar of it is retail media, like commerce, shopping. And that's something we definitely want to get into, but small startup, fine amount of time, that's more future facing. What's

the biggest opportunity in ad tech as it relates to generative AI? So

again, I think we can break this down into sort of the platform shift versus the technological opportunity, because both of those are big opportunities in very different ways. So in terms of the platform shift, that's where you see sort of the first cracks in Google search market share, for example, sort of this question of how are people going to be aggregating and pulling for information, answer engines, generative experiences. I think no one has a solid answer on what the future of that's going to look like. But even seeing 3% market share dinged from Google in the super early days, it suggests it's going to be massive. And then separately, there's the technological opportunity, where I think you definitely see what the Vanderhoeks are doing with buy-in. And there's a ton of opportunity just to automate and add efficiency to sort of the existing layers of the ad stack. But separately, I think the question that's worth asking and something that has here to for been underexplored is what is possible now that wasn't possible before. And that I think is a big part of like what we're here to talk about today is what can we do now that we couldn't do before, whether it's in ad targeting and creative generation, personalization, optimization, in terms of sort of really the question around what is an ad and what can an ad be?

Yeah, I mean, I think the platform dimension of this itself will be a giant business, right? Like the global search advertising market is, I think, over $300 billion a year. I think Google alone does like 160 or something like that. So one could certainly imagine that the open AI or perplexity businesses combined will be of a similar scale eventually. If they're not, you guys tell me why. No,

I think it's viable. And the way that sort of we think about this, like putting it into historical perspective is ad tech is exciting when there are platform shifts, whether it's the dawn of mobile, the dawn of CTV. The really big ad tech companies and the big exits are born sort of at that cusp of a platform shift where the incumbents are struggling with the innovators to live. They have their billions of revenue walked up in executing what they've been used to executing. And so they're not able to move as fast to adjust to a new format where the old tech doesn't work as well or at all. And there is an opportunity for new players to play in the space. And what makes this particularly exciting is it's not just a platform shift, it's a technology shift. And that's something where we haven't seen those happen at the same time, probably since the double-click era. I think that's

right. This is an audio podcast, but i'm smiling because i don't follow a lot of people on twitter michael you and joe you're both uh in my in my follow list um and michael you say a lot of cool stuff a lot of stuff that i don't understand but once in a while you you drop a banger ad tech is exciting during platform shifts is an absolute banger so that might even be the title of this episode. Do we think that there's just a direct line for every dollar Google loses, some LLM is going to gain? Or is the fundamental nature of getting an answer versus doing a search where the opportunity to serve a whole lot of stuff, including a whole lot of ads. Does this mean that getting answers, right? Because I don't mean to call it search anymore, is going to mean a different business model and one that might not be just as easy to build, you know, such a monolith and giant business that Google did. Does that make sense?

Yes. And I think spitting off the cuff on wild speculation, I think it absolutely changes. And I think one of the signs you can look at here is the fact that publishers are suing perplexity and suing open AI. It's sort of this problem of the zero-click search dilemma, where the less friction there is between a user and pulling for information, the less opportunity there is for advertising. And the less opportunity is for those publishers to monetize via display ads on their sites. And you already see startups popping up to deal with bits and pieces of this, like Tolbit, Prorata. And I think we actually wrote up a big market manifesto around what it would take for an industry collaboration for revenue sharing and attribution between publishers and aggregation engines is how we were referring to it. And I think it was definitely premature. It broke Ari Paparo's brain a little bit to try to read some of it. But that's the future of where things are going, I think, is that sort of the world where Google existed in sort of a handshake agreement between publishers. We'll give you traffic, you monetize via ads. And really the like answer engine play is almost backstepping to say, we are changing the terms, pray that we don't alter them further. And you get these AI startups being sued because the publishers exist at Google's whimsy. It is their lifeline. They can't sue Google. But I do think it's precedent setting for this idea of duplicating content or effectively duplicating content such that it removes the need to go to an underlying publisher site, I think that being unmonetized and not revenue shared with the publisher is not going to last. I

think there are two dimensions to this in terms of the potential disparity between AI and search as we've known it over the last 10 to 20 years. So one is sort of univocal answers versus a sort of a polyphenous array of possibilities, right? So it's like, just tell me what it is. Like, give me the one thing. As you said, Michael, like that doesn't lend itself to as much search surface area as like, and here are the 10 possible things you might want to know or look at, which also has like devastating consequences for display revenue, because, again, people aren't going to be navigating to all those different properties. There's a deeper problem, I think, which is that if you transition even from, like right now, people are using AI, sort of like the next generation of search where they're like, give me the answer. But you could go beyond that from give me the answer, which still involves a moment where the various ads can be displayed to just do it for me, just make the decision on my behalf. And when you get to that point, which is like the real like AI agent paradigm, then I really see the trouble with ads because then you're no longer having the user sort of put text in and like examine an array of options. They're just having the AI do the thing. And then the only opportunity for the advertising would be in potentially reaching the agent itself, which is like a whole other debate that's going on right now. Exactly.

And I think that's something that is definitely going to happen, like very much so in like online shopping, e-commerce, in like pricing, personalized discounts, in like really, like it almost becomes more of this like real time, like bidding-esque optimization problem of like, if you're pulling for information, okay, here's a sponsored opportunity what makes it actually good for the user then like are they offering a better discount are they offering something more suited to the user preferences like i think that that's sort of like agent to agent quote-unquote advertising is almost not even advertising as we traditionally conceive of it like i don't think ads are going away by any means there will always be a surface area to put advertising. But some of how things have been monetized previously is very likely to shift more to this sense of price discovery, price optimization, and personalization. And that is very exciting. And I would say a little parallel to what we're working on directly right now, other than under the hood in terms of like existing advertiser selection and optimization for click-through

rates and ROAS. Well, and we also know Silicon Valley has a huge bias against ads and for SaaS subscription-type models. And so I wonder if they can have the AI agent and they're like, look, we're going to try to charge people a SaaS-type fee for an agent over like the information being free and then like generating revenue in the form of ads. I imagine they would go for that. But that gets into like a deeper conversation about what that I'd like to ask you to comment on Michael in terms of as someone like operating at the intersection of like Silicon Valley AI ecosystem in the New York ad tech ecosystem, what are some of the tensions you're seeing there in terms of how this will play out?

Absolutely. And I do think that Silicon Valley has broadly an ideological, almost irrational distaste for advertising. What we've heard from a bunch of prospective clients is like, oh, wow, I didn't realize ads didn't have to suck. And that's a big part of what we're trying to do here is almost evangelize for the idea that you can make better ads. I think in terms of user behavior, you can look at, for example, the fall of Neva. Some things users are used to not paying for. And even in cases where there are subscription monetizations, I think the evolutionary line of that is towards freemium models. Netflix, for example, started out subscription only, and they eventually introduced an ad tier. That's just a matter of it's a financial optimization problem where most services that are purely subscription would probably be better off freemium. And it's a matter of at what point in the growth and scaling curve, like, is that actually the optimal play? For

those unfamiliar, Neva was a search engine, you know, essentially like an alternative to Google started by a former, believe it or not, Google ad exec, Srinhar Ramaswamy. And ultimately, it was acquired and shut down by Snowake um and strunar is now the ceo of snowflake so michael to your point an alternative to google that is freemium or paid that's not the big idea no

i think the big idea here i would say is almost that we are agnostic to what form the internet takes six months from now two years from now we think it's absolutely going to change and sort of the like principles that are definitely going to be happening are sort of, if you look at the dynamics of how LMs work, how like human language input works, how we are measuring like new forms of context and ad selection, that's sort of the consistent factor is it's not necessarily just about web pages, but about like a different kind of input and output, and how do you make ads relevant in that? So whether this ends up being perplexity, Google Answers, etc., are sort of the final form of that, or if there are things that we just haven't imagined yet, we are very unopinionated about what that looks like. Either way, it's going to need a monetization layer, and

it's going to be ads to some degree. You are an ad tech guy. You formerly worked at Moat before doing this. Correct me if I'm wrong.

Correct. I was there from relatively early days up through the Oracle acquisition. Got it. What'd you do after that? Took a year soul searching, traveling around and found my co-founder. Grew your hair long? Grew my hair long, ad tech Jesus a little bit and found my co-founder in an alleyway in stonetown zanzibar we started talking about his background in political tech my background in ad tech and we sort of just started jamming after that like uh we launched a small political tech startup leading up to 2020 it went disastrously we learned a ton we decided to do something that actually made money after that and uh like move towards e-commerce and towards i would say a very similar idea to what is still fueling what we're doing now. This optimization problem of incentives drive what content gets made and how users experience the Internet. And that is the ideological strain that I have kept from the mode days is this laser focus on what you measure drives what outcomes you get. And with the context of us being at a platform shift, being at a technology shift, and looking at the history of, Google was very aware of the tension between search and ads. We look at this very similarly of, okay, let's be very thoughtful, very considerate of how do we make sure that five years from now, 10 years from now, we don't have sort of this proliferation of a slop engine, because that's what is financially optimal for people to make and for people to game the system. You're such an interesting dude.