Scope3 COO Anne Coghlan on the AdCP Launch and Agentic Advertising

AdLib is the DSP agnostic buying platform to scale your programmatic campaigns smarter, faster, and more profitably. AdLib helps independent agencies and marketers deliver premium programmatic performance without the complexity cost or overhead of traditional DSPs. Their platform provides the tools, data, and automation to run enterprise-quality campaigns from CTV and audio to geofencing and YouTube all under one roof with zero trafficking headaches. Plus, they're not just a platform, they're your growth partner. The AdLib team helps you win new business, expand budgets, and deliver measurable ROI that keeps clients coming back. It's time to unlock programmatic. Join leading marketers worldwide who are using AdLib to deliver more campaigns with better results. Check out GetAdLib.com. Welcome to Open Market, a podcast about building companies in the advertising industry. I'm Joe Zappa, I'm joined by Eric Franchi and Anne Coglin, COO of Scope3. How's it going, Anne? Hi, guys. Yeah, I'm very well. How are you both? Good. You've been having a very busy week. It is a busy week, and it's end of day here in the UK, so it's been a very packed week for me, for sure. For those listening, this will probably hit in the next week or so. We're recording this on Friday, October 17th. This is the week that the announcement of AdCP hit the industry ticket by storm. We wanted to dive into it with Anne being from Scope3 and one of the organizers and leaders of the movement about everything. We should say that my fund is an investor in Scope3, and Scope3 is a client of Joe's business. With that, Joe, where do we start? Anne, for those who live under a rock, what is AdCP? Why should the people care? AdCP, it's like in between a three-letter acronym and a four-letter acronym because the D is small, but the ADCP is the ad context protocol. So this is something that we launched on, I guess, Wednesday, the 15th of October as an industry-wide collaboration with over 20 companies. And it is a new protocol designed to provide the foundations of how agents should talk to each other and communicate as we enter into this new era of agentic advertising. And we figure out how agents can help hopefully solve a lot of the problems with the existing paradigms of media buying and selling. And Scope3 has launched a couple of agentic advertising products of its own over the past year. So how does the adCP launch relate to your own products and vision? Yeah. So in March of this year, I nearly said last year, it feels like a long year, we launched the idea of brand agents. And so this was the realization that if you could, as a brand, put all of your operating instructions as to how you wanted to buy media, essentially, give those operating instructions to an agent and the agent could essentially do your bidding, then you could find inventory that was more effective. You could deliver better outcomes. You could make brand safety and suitability concerns just be part of the fabric of buying because your agent would know what to do in any given scenario. But there needs to be a common language of an agent talking to another agent. There is examples of this. So the ad context protocol is based on MCP, which is the model context protocol, the Anthropic, the parent company of Claude, who's my favorite. But we kind of got together with some other companies within the ecosystem are also thinking about agents to figure out, you know, if we will have agents, how are these agents going to talk to each other and how are we going to do that in a way that means that we can have a flourishing, open ecosystem of agents are not fighting about integrations. And so another way of thinking about the ad context protocol is that it's a universal ads API, essentially, that enables this to work at scale. Talk about some of the other companies involved and maybe even take us behind the scenes of, like, how this all came together. Because it wasn't until the webinar kicked off and Brian's email that I think we found out that there was 20 companies, which is like awesome. And then on the webinar itself, the scope three buying agent executed a buy with swivel being the sell side agent. Talk about the company is like what like how this all kind of came together and then, you know, how these buy and sell side put together. So with all good things in this day and age, there is a slack channel, all of these things. So there are a small group of founding partners, and I've got the list so that I don't forget any of the any of them. So then the founding partners are Yahoo, scope three, pubmatic, swivel, Triton, digital. I actually don't have the list. I lied. I'm going to get it. Make sure I've got everyone. M loves M loves all her children equally. Of course, it's optimal and a big team is joining us as a founding member as well. So they weren't announced on the press, but they are a founding member now. And then a big list of what I'm going to not going to list all of the large members, but you know, a big list of people coming together. And it's it's really the builder community and the geeks that make the glue of the ecosystem where you have people discussing, you know, the right ways of actually setting up the protocol, how we should split things out, different scenarios, what fields we should have, what the actual interactions should be. So an incredible number of technical individuals across all these companies being extremely collaborative, essentially. One of those is kind of almost feels like you're working internally where you're trying to problem solve and you're disagreeing. So that's one side. And then the other side is figuring out, you know, what this body is, what this consortium actually turns into. And that's been happening over the summer. And then figuring out how we we launched this and the marketing teams of multiple companies, including scope three involved in putting on that. That webinar that we showed on Wednesday, which is open and available on YouTube now. Why the evolution from programmatic to agents in the first place? Like, you could imagine it's more efficient for the buyer, you don't need as much technical or platform specific knowledge. You could imagine it's a question of the signal that's getting communicated and better being able to understand what's the information in terms of content and audiences on the sell side. Give us sort of the case for the agentic evolution. If you're a brand, I think what you really want is to actually buy ad space to find your people, to find your consumers or potential consumers. So the best way in a sense of doing that, if you think about the direct I forget programmatic for a second, you think about the direct I/O paradigm. You kind of have a really good insight of where your consumers are, what they're consuming. And you go and you execute a direct buying relationship with a publisher that is reputable. So, you know, there's high quality. You're not worried that you're being scammed because you've called the person or I guess you've emailed them and you are able to see that delivery and tie that well to outcomes. So you're essentially putting your creative into the publishers ad server and you know that it's going to run and you've kind of organized that. That is the scale at all, unless the publisher has an API that you can connect with. But fundamentally, you're not going to be able to, as a buyer, do that. And you're not going to be able to have engineering resources and connectivity and assuming every single publisher that you want to work with even can do that because of this, you know, how big and how strapped publishers could be. And so that doesn't work from a scale perspective, but you do as a brand get to find ways of being really creative in actually pulling in your consumer because there is all the things that a publisher can do from a fun, native, rich experience perspective for creatives and you flip to a programmatic, you lose a lot of that because you can't be super, super custom across the entire way of thinking about the open internet. There is outcomes based buying on wall gardens and social platforms. You're probably going to be doing that. And then you that's one alternative and then you have the programmatic pipes where you lose a lot of the richness of even of the signal as you get to the Nexus. And I, this is my favorite word and I know that our Nexus has conversations there. But I'm just going to say Nexus anyway, the Nexus of the DSP essentially, which is the decisioning. So there in programmatic, you have the scale, you lose some of the richness in trying to find your people because you just can't activate against all of the signal. So thinking about those two, two worlds or three worlds, if you think about the API example as well, we realize that instead of translating a really rich brief of what a marketer wants and it not going well because of needing to squash it into the programmatic pipes. What if the buyer agent could actually just talk and have that nuance discussion at scale across all of the different representatives of publishers? You can just think about that as a sales agent. So for anyone that was at the pre bid summit in New York the day before the launch, I wasn't there. But I saw a video recording of my co founder, Brian, who actually literally got people up on stage and said, you be a sales agent. You be a buyer agent, have a chat, go essentially like go to one side and talk about the inventory that you have available as a sales agent. And you explain to the buyer whether that will meet their requirements. So yes, you can have a dialogue and a conversation. That's what we see sales agents doing. The important part of that is, I don't want to say offline, but that's not part of the how we would think in the programmatic world of our minds automatically going to an auction. And I think that is one of the things I'm seeing as a misconception with our CP is that this is like, oh, my gosh, you're you're you're having an agent decide on every single impression whether or not to buy. And like, no, that is actually crazy from a speed perspective. Doesn't seem like it's a good idea from an environmental perspective that would suck. Actually, we don't need that anymore. We don't need to be thinking about the impression. So we don't need an auction. We can just have essentially a conversation that then means that there is a direct purchase of inventory that goes into the publishers ad server. They can be some real time elements as well, but it's a different way of thinking about it that is just separate from programmatic. I'll let Eric come in, but I would just summarize as the shift to agents unleashes the benefits of direct ad buys at infinite scale. I can't take credit for this quote because I saw it on LinkedIn this morning. But someone commented that the other benefit that they saw from our CP was that the industry's black box just became glass, which I think is incredible. Because some of the thinking about how these buyer and seller agents will communicate to each other, it's direct. It's a direct communication. And so there are no intermediaries. You can have a representative of publishers who would be a programmatic sales agent, for instance, who would be representing publishers. But it's still then directly talking to the buyer agent. You solve essentially the problem of a lack of transparency. So I just thought of that quote and I need to have everyone aside this guy. But he said he said that the industry black box just became glass. I thought it's incredible. Let's go back to one thing you said about outcomes. Wall gardens are now becoming synonymous with outcomes based buying and driving outcomes. And one of the rubs today against the open internet quote unquote is that it hasn't truly been a place for outcomes based buying for a variety of reasons. And Terry Colasia talked about this on his part of the webinar, which again, must see of how AI can be the shot finally for the non-wall gardens to compete in an outcomes environment. Do you agree with that? And do you think things like ADCP help to really accelerate the potential shifts to outcomes for the open internet? Well, I think one of the things I've already said is that we don't have to talk about impressions now. We're actually talking about what you want to achieve and measure against that. So actually within the protocol, you can define what your outcome is. You don't have to buy on impressions anymore. That is a huge deal. But to do that, how do you facilitate that with the shift agents? Like wouldn't you need to build all sorts of measurement stuff on the back end? Or is it that because of ADCP, sellers and retailers and whatever can communicate with each other so then that conversion data is being made available or something like that? So because it's the creative as well, there's still all of the existing infrastructures like your conversions API, we fire pixels, all of that stuff still exist and is the same. But within the protocol as well, like the communication of how you need to kind of report back. But it's then published ad server data, talking to a creative ad server via data that can be connected and there's not like 12 different systems to try and figure it out. Yeah. So the contention is an agent communication system should obviously make buying more efficient, should increase trust again, benefits of direct ad buys at an infinite scale. But it should also improve like the performance measurement loop so that you can iterate more quickly. And because it's agents doing it, ultimately not humans, you would imagine that that performance oriented iteration would happen more quickly. Yes. I would add that as humanity, I don't think we're fully there on fully automated agents yet. And so there has to be human in the loop checks in the same way that, you know, 10 years ago or longer when we were thinking about a programmatic, you know, the first kind of machine learning algorithms in optimization bidding. We would have controls about Minz and Max Bids and we would be doing lots of human checks. Here, there are checks built in where you could have the human like approving the buy or a human double checking that the creative is correct before it gets added into the ad server. And so a lot of what we showed in the demo, I would say a little bit more human than fully agentic in terms of the agents just making all of the decisions because we wanted to show that there is the ability to have like set change essentially as people get comfortable and confident with it. But everything else, I agree with what you said, Joe. The day before the webinar was a pre-bid conference and TTD was on stage talking about open ads, which is a new auction wrapper that they say will make existing programmatic buying fare, the pre-bid fork. What you were part of with adCP is much different. It might be a fork of programmatic or the next gen of programmatic. Do these things kind of go together? Are they intention? Is adCP just going to like hop over open ads? Oh, so many great questions there. So I think that the. I've been waiting for this conversation. Yeah, that's a list of questions that I just firing. I want you to know that if we did YouTube style headlines for this podcast, there'd be like Scope Threes and Coglin eviscerates TTD's open ads called Jack Free the Dinosaur. People don't clip that. Go ahead. Well, like the what's the premise of that? But Tradesca saying we want we want auctions to be fairer. And that's the way that the open internet will flourish is a fair auction. And there's a lot of fighting about who owns the auction and is it the buy side or is it the sell side? And are you a DSP or are you an SSP when everyone funds, mentally, is an ad network and Tradesca like building publisher tools, but we're a DSP because that's in service of the of the buyer. Still, even though we're building stuff for publishers. But fundamentally, it's all around the idea of the auction. And that's like the open RTV spec is how all of these constituents come together. What fields get past? How do we know that things are as transparent as possible? But at CP, you don't need to have an auction. It's just a different way of thinking about things. And so Tradesca trying to solve an existing problem that they have. And we are as a group of, you know, almost 30 companies at this point. We've gone from 200 people in the Slack channel on Wednesday morning to 405 people in the Slack channel. And I checked just before this call and that's two days, double in two days. This is a movement. This isn't just a little idea of a new protocol. This is a way of thinking about how can we see the existing ecosystem and not try and solve little parts of it? Is there something that we can do that's just entirely new and cool, but also leveraging the future and bringing it into the advertising ecosystem? Because I don't know about you guys, but even not in the world of advertising, everyone's talking about AI. So there's definitely a lot of appetite for innovation. I think, you know, the answer is these things are in tension and potentially like if agentic advertising were to really come to fruition and achieve its full potential, it would represent the next evolution of the ecosystem. And I think, you know, maybe make the open market obsolete. Obviously, we're a long way from that. A lot would need to happen for that to come to pass to which point. And in talking to a lot of CEOs in the ecosystem about this, which I, you know, I spoke to all of my clients about in the past few days. There is skepticism about like, OK, but how are we actually going to get publishers who are often thought of as fairly change resistant to to adopt this and to get comfortable with, you know, agents selling their ads? Even though, of course, there will be, you know, a human in the loop and it's not going to be like a total one to one replacement. Who are some of the companies that are jumping in on this? And what do you think are the keys to adoption? So Yahoo have built a sales agent. They're a company that has a really strong, I guess, elite sales, like human sales team. But that's not something that can scale to them. That's why they sell their inventory programmatically. There are, I think, three at least three or four other publishers that are in the launch group. And then you have companies like Swivel who are building sales agents on behalf of customers. There is an open source sales agent that you can implement. We've had a couple of people internally at scope three being like, I tried to build my own like my own sales agent on behalf of my blog to see how easy it would be. And so, you know, I think there's always going to be a, you know, cost benefit analysis and the start of a new ecosystem, start the marketplace. Basically, there's always going to be this inherent tension between the demand side saying, hey, like, I want to buy some stuff. Which sales agents can my buyer agent talk to? And then the same on the other side. But, you know, I think it's very early days. We've seen the first agentic by and there is the barrier to entry is a lot lower than changing ad server. But I think there's also, you know, you saw what happened when pre bid was introduced and it was solving a real problem. And it just became like every single publisher in the world had pre-installed within it. Might this be where the SSPs go? Because if you think about the origin of the SSPs, they were at first, they kind of like organized all of the intermediary demand for publishers, right? Like AdMeld and PubMatic and Rubicon. When they first started, they just like created this waterfall of demand. And then they evolved to bring all of the programmatic tools to the publisher space to keep up with the buy side demand. Might this be the next evolution of what we think about today as SSPs? Maybe tomorrow it's going to be, you know, a different set of companies. Yeah, it's a really good question. So there's a bit of a debate happening within the slack channel of the working group around. There's a couple of diagrams floating around of the protocol and then boxes with like, this is like the DSP and this is like the SSP. And it's trying to fit the existing labels onto the new ecosystem. And I joked about ad-cp being a four letter kind of a three mark acronym. Like, if you think about everyone being a network or everyone kind of being a platform in a way, companies will decide how they want to play. And, you know, if you've got specialty in representing supply, then maybe, yeah, maybe you build sales agents. If you have specialty representing demand, maybe you build buyer agents. But will we be talking about all the Loom Escape boxes in five years time? I don't know. No way. I don't think so. I agree with you. And how do you distinguish between what we might call a true agent and like general LLM functionality? So, for example, like, it's already very useful to shift programmatic from like the very complex DSPs, which are sometimes described as like the cockpit of a 747 to, you know, natural language functionality, where you can just say, do this and it does something or it gives you answers. Is there, you know, any whitespace between that and then an agent that is really like semi-autonomous and doing a lot of actions for you? And where are we on that continuum? Yeah, it's a great question. So there's kind of two labels, right? There's traditional AI, which I think is quite a funny label. And then there's agentic AI. The way that I like to think about this and explain this in a relatively, hopefully straightforward way is I think of traditional AI as like my career coach. So they don't actually make me do anything. I just ask them stuff and then I take it on board and then I'll do it myself. So, you know, if you're talking to Claude or ChatGPT or Brock or Gemini, I think I've got many months, you ask us and stuff, you get it, do things, you'll say, cool, everything looks up the way. It's not agentic because it's not planning, it's not reasoning. It's not doing things almost like before you've asked. And so then I think about agentic AI as being right now like my amazing intern, who I'm going to give very like clear guardrails to because they're not going to be answering emails and coming on podcasts on my behalf. But they're going to be given a very specific set of instructions. But as they get more mature in their experience, they're going to be making calls and doing things and planning and executing without me knowing every single thing that they've done. So that's how I think about the difference. There is a kind of a planning thing, step within agentic that's different to the traditional AI. So that's how I think about the two different scenarios. Makes a ton of sense. Before we get into a little bit, it was specific to scope three. I want to ask you a question in, how has your work changed with the advent of AI, but then also you being so deep in bringing an agentic product to the marketplace? Are you using AI more or using it differently? Do you have anything you can share with us in terms of just like tips from a work style perspective? It's not even just work. It's in my personal life as well. So I have a Claude agent that helps me do QA tickets on our product. So instead of going into Notion, which we use as our ticketing system, if I find a problem, I have a little project within Claude and I'll basically be like, this is the problem, create a ticket in the right place and it goes in and it finds the right database in Notion and just does all that for me. I don't have to worry about kicking the drop downs and saying all the stuff. It just does it. I use willow voice. So I talk a lot, which is great. Which is why I'm glad that I work remotely and I'm not in an office because it's very annoying when I'm in a wee work and I can't talk to my computer all the time. In my personal life, I have a, my son is that there's a kind of an annual class representative that is the parent that represents like all the other parents and no one ever wants to do it. So this year I said that I would do it, which is kind of silly because my job is very busy. But I basically created an agent which reads my email, will pass out all of the PDFs and all of the emails that I get from the school and then will email me back and craft the messages I need to send to the parents of all the kids in the two languages. My son goes to a bilingual school, so in the two languages that's required. So that saves me an incredible amount of time. I get all the reminders essentially. So my agent is the class rep agent, essentially. So that's pretty good. And then, you know, we're continuously figuring out how we can be more effective with our own internal use of AI. So I use a little bit of, I'm not an engineer. I don't code, but I'm using a little bit of core conductor to do some PRs or approve some PRs, mainly around like copy and documentation, tons of stuff. Our marketing team is amazing. They have, I think it's called, thank goodness, it's AI Thursdays or something. They create stuff. I think we've got butlers that, you know, proofread things and team connections as we're remote around, you know, asking each other questions that can be fun. So just, yeah, trying to embed it in everything that we do. And then of course, actually within the product itself, we're laser focused on a really good experience for people as they create their brand agents. So anyone who's achieved anything and is doing anything provocative, and especially anyone who drums up as much attention as Scope3 does, and Brian does in particular, generates haters. And what the Scope3 haters say is you guys were doing sustainability during, you know, peak woke, and then it wasn't as cool anymore. And now you're doing agents because that's what's cool now. And essentially you're jumping on a bandwagon. How do you view the company's evolution? I think even in the first year of Scope3, we were trying to examine the existing ecosystem, which was the one that Brian and I know very well, programmatic, and trying to poke holes or find a way of seeing it in a new way that would actually get people to make the systemic change that we wanted. We had like economic theory about, you know, like carbon emissions being the kind of tip of the spit, the wedge in by which we could then solve this problem of ineffectiveness, bad outcomes, supply chain, basically just getting rid of all of the crap. And I think, you know, in last year when we realized that there was just a gap in how the industry was thinking about measuring, you know, we've been massively part of the GMSF, the framework for measuring carbon emissions in the GMSF right now. There's still nothing about AI. And so you can think about looking at the existing supply chain, measuring really well the carbon emissions there. But all of the AI, can you imagine AI sitting on top of every single one of the companies that are part of these large, not effective supply chains? Like that's a big problem. And so we started measuring that. And it's very early in the hype cycle for AI in general. And when you're in the kind of moving towards this peak of inflated expectations in the Gartner hype cycle, people are not going to be moved. You don't want to not do something because you don't want to be left behind, especially when it's from an environmental perspective. You know, there's a correlation between cost and energy, of course. But we realized that there was a huge risk in the existing programmatic ecosystem, just having AI sprinkled on top and it's just becoming much, much worse. And so seeing and, you know, what we do a lot is be very curious about emerging technology, thinking about not just existing systems, but potentially manifesting new systems. Is there actually like, is there actually a way that we should be thinking bigger and zooming out more and not just trying to fix what's there, but actually be like, actually, maybe there's something even more powerful. And so the one thing I think from a sustainability perspective is computes come down and efficiency has gone up. But what I'm talking about here is not auctioning every single impression. We're talking about agents communicating in a more finite way that kind of brings the sustainability goals still being there. And we measure our carbon, we have it in our transactions. It's part of the goals that we have to, you know, tell our board about. It's part of our DNA and why so many people joined the company. So it's something that we think a lot about, I think it's just less direct now. Yeah, I think also I'm biased. Obviously, I work with you guys, I like you guys, but I think people are envious of the amount of attention you guys are able to orchestrate and Brian's presence in particular. So there, you know, there are some, you know, motivated critiques. But I think there's also just nothing wrong like, and you give a very cogent explanation, right, in very detail. But there's also nothing wrong with being like, well, we started the company in what, late 2021, early 2022, and then the most significant technology to hit the market since the launch of the smartphone in 2007 arrived in the explosion of chat UBT usage in late 2022. And any startup that had launched, you know, a year before the explosion of consumer usage of AI would have reevaluated its product roadmap, its orientation and its approach, given the advent of that technology. And there's nothing wrong with, you know, AI emerging and saying, oh, this is going to change everything, which is a very honest realization that virtually everyone in the tech industry has had over the last few years. And then saying, okay, how does this shift our mission? And maybe sustainability is still a component of it, but maybe actually we're going to reinvent the entire advertising ecosystem. And that's going to be a much bigger endeavor of which sustainability is a component. Absolutely. Definitely still a component, but it's part of a bigger thing. Right. We talk a lot about marketing on this pod for obvious reasons. And I think, as I said, scope three generally crushes it at marketing. You guys have a great team. You're able to generate a ton of attention, which is the first thing you need to do to make marketing effective. If no one's thinking about you, you can't accomplish anything. It doesn't matter how good the message is. And I think that was probably central to the success of this launch. My understanding is you've got nearly 2,000 people to sign up for a webinar at 11 a.m. on a Wednesday, which is pretty insane. Because there's a lot of talk about webinar fatigue and really the whole industry tuned in for this. How did you guys make that happen? How do you make the launch so successful? And how big a role does the attention you guys are able to drive play in that success? Yeah. And the first thing was that the majority of people also attended. It wasn't just that more than 2,000, almost 2,000 people signed up and like 50 people joined. So that was what was really incredible about it. I mean, there were over 20 companies as part of the working group and as part of the launch. And it was, I would say, not the best kept secret either. And so I think that also helped a little bit. So it wasn't, you know, it wasn't just a scope three effort to drive the webinar push. So that's kind of the first thing. So when you have multiple companies involved, if you take a slice, then that's helpful. But I think it goes back to storytelling a lot. And I would say inception and interest as part of the storytelling. We've been talking about a genetic advertising for months. But I think what Ryan has done really well is he's been inviting people into the conversation as part of that storytelling. And so, you know, that's not a like a long post. And then it's like, what do you think at the end? Which, you know, I think a lot of people use AI to generate those. It's actually a genuine interest in learning about what people's opinions are and collaborating. And that translates into the actual, you know, working groups and the consortium that we have as well. And so I think Brian's been using his platform to learn and ask questions. And that makes people also feel like they're part of this as well. Because A, because they are. But B, you know, we've had people comment on Brian's LinkedIn post. That comment goes into the working group. It's like, this is a great idea. We should do something like this. So people are actually involved in participating to the ad context protocol without realizing it. And that's really powerful because then you almost kind of see and resonate with it more because you have been on the journey even if you haven't known about it. And now everyone knows about it. So I think that's something that's, I mean, it's part of the reason why I love working with Brian is he's endlessly curious and will always be trying to refine his thinking and improve his thinking and kind of sometimes change his mind. But it's all about bringing people in. And I think that's why the launch and let's hope the continuation is really, really successful because just the level of, so like shields down, kind of not being afraid to be told, no, I think that's a bad idea or why don't we do it this way, is really high. And so that's kind of, I think a lot of the success of the webinar and the attendance and all of that is because it doesn't feel like it's not scope threes thing. It actually feels like something that is for the industry. And that's what Amanda, who did the initial kind of talk at the beginning of the webinar set. It's like, it's for the industry, it's for everyone. We're going to improve it. There's probably things that we've missed, but we're really excited to have a significant number of participants join and be involved in it. Yeah, it was super well done. Do you have any personal predictions for what the short and long-term trajectory of, I don't know what we call it, right, like an agentic ad text might look like, like, you know, do you have a timeline for when, you know, we start to see real commercial impact of agents? Do you have a timeline for when buying is going to be mostly agentic? Like, you clearly think about this stuff. Do you have any like thoughts around, around the ultimate trajectory of this? So at a, a working lunch with a group of publishers a few weeks ago about, and this was kind of an introduction, like, what do you guys think about this? Like, what are we missing? What, you know, do you like the idea of sales agents? I think it was about two thirds of the way through the lunch. Someone asked us this question. And someone said, when do you think agentic advertising will be the de facto? And then everything else will be kind of, you know, less. And that was the moment where I was like, oh, huh, it's a when. It's not an if. And so I think we don't know how, how quickly this will move. Certainly, you know, there's, there's a bit of recency bias and excitement in the fact that the webinar was a year ago and we're seeing all of this momentum. But at the same time, if you look at another industry movement, which was, you know, trying to solve a different problem, but still trying to reinvent or reinvigorate some of the issues that were or solve some of the issues that were existing. And that was pre-bid a long time ago. That adoption was fast and a lot faster than people anticipated. And then if you just look at the, the pace of change in general, in, you know, how people are adopting new technology. I mean, did anyone say agent a year ago? I mean, the amount of development, even on the frontier models and the, the releases that are coming out on a very regular basis. You know, I think it was, it was nine months ago when Sam Altman said, "I don't think I'll ever see ads and open AI being the same sentence." And now, now it's like, okay, so when's that going to happen? It's just moving, I think, quicker than we can envisage. And so what the most important thing is that we get to standardization. Like everyone kind of kind of agree on the protocol and kind of adopt it quickly so that we're not ending up with fragments and schisms on it. And so I would say I'm pretty optimistic about what this will look like. But I think it's too early to give kind of numbers and inflection points and dates in the sand. Well, I think it's 6.17 in London. And we should probably let you get to your weekend activities and hang out with the fam. So, and thank you so much for coming on. Thanks guys. It was great to, great to chat through this. This is awesome. Thank you so much. And we learned a lot. And we will see you next week, everybody.