Ep 29. Agentic Marketing Is Here - But Can AI Actually Stay On Brand? with Kashish Gupta, Hightouch

Speaker A: Welcome to another episode of the AI Edge podcast. We are here to help marketing and advertising professionals stay on top of the AI news and what it all means for our industry. Uh, all right, we're here for another episode. We have a really exciting guest with us today who's been in the news a lot lately. Uh, and so we are going to make some news with him, hopefully here. Kashish Gupta. Uh, uh, Kashish, I'll hand it over to you to say quick few words about who you are and what you do.

Speaker B: Sounds great. Uh, thanks for having me. Kashish Gupta, uh, co founder and co CEO of a company called hitouch. And what we've built is the agentic marketing platform. We work with some of the largest B2C brands, folks like Domino's Pizza, Spotify and other large brands, um, to help them really just do their marketing end to end in an AI native way. And so looking forward to just jamming on all things Martech and Adtech. There's a lot changing right now in the industry, and hopefully the three of us can help clarify some things. Awesome.

Speaker A: Ah, we're thrilled to have you. And you said three of us, which means there's somebody else here. Miles Younger. Hi, Miles.

Speaker C: Hello. Hello.

Speaker A: How are you doing? Um, all right, let's jump into top AI news of the week. So in our newsletter, uh, we made a pun, because there's just unlimited amounts of puns that you can make with the city name. Can. So can we talk about AI this week?

Speaker C: Do it.

Speaker A: Let's jump in. Um, so there was obviously a lot of announcements. It was the Cannes Festival of Creativity in the south of France. A lot of people there, a lot of partnerships, announced new product launches, et cetera. We do not have time and energy or the knowledge to break down every single one of them because there were too many. But we'll try to hit on some of the top ones and get, uh, some hot takes on them. So to start, Amazon, uh, launched, uh, something called Alexa agentic ads. And, you know, Amazon. So, so to break this down, what this is, is, uh, everyone is familiar with Alexa. It's the voice, you know, activated assistant, uh, from Amazon, Uh, and now, you know, I think as of a few months ago, they were enabling ads in the kind of conversation back and forth with Alexa. So, like, if you ask for toilet paper, like, they would recommend maybe a particular type of toilet paper, uh, based on, like, who sponsored it or not. So that was already starting to happen. But now with Alexa plus agentic ads, they are, um, step Further, and they're using the data, uh, to basically like change the shopping experience and allow you to natively even check out through Alexa. Um, and so, and there's a lot here, like Amazon had a bunch of other announcements too. Uh, not all tied to can, but there was another announcement that I thought was interesting, which was Amazon is now enabling shoppable ad formats, uh, to go across the web on third party publishers, which is kind of like AdSense in a way, but it's Amazon's ad units and it's shoppable advertising as opposed to like Google AdWords and AdSense kind of melded together. So, um, lots of stuff coming out of Amazon. Uh, Amazon I think over the last few years is just super duper accelerated monetization of every single surface they could possibly have through ads. It started with Amazon Prime Video and now obviously it's Alexa, it's shoppable ads, et cetera, et cetera. Um, so any quick takes on this, what it all means? Uh, Miles, we could start with you on this one. Uh, what are your thoughts?

Speaker C: If I'm going to order toilet paper agentically from anyone, it's going to be through my A L E X A. I can't say the name or activate her. Um, uh, and I mean that's, that's a case where I can absolutely see myself shopping that way. And just like, you know, even if, like the price doesn't have to be perfect, like the brand doesn't be perfect, the sheer convenience factor of me being able to just do that spur of the moment without having to like open my phone, pop open a laptop. Uh, so if anybody can pull that off, it's Amazon. I actually had like a side question for you Kashish, is does high touch, uh, deal with, um, audio at all? Is like one of your, the end points that your customers are using for marketing activation or customer relationships?

Speaker B: Not yet, no.

Speaker C: Okay. All right. I was curious if you had any inside information on how people are using audio these days? Because that's the part that I'm just like, I don't know, is this really going to be a thing or is this a new habit that people are forming? Basically, yeah.

Speaker B: And I think.

Speaker A: Yeah, sorry, Kashish, go ahead.

Speaker B: I was just going to comment. Um, a lot of the folks that are working on audio still tell me that it's not fast enough to that real time computation. Right. Like what happens is that the agent has to wait for you to finish talking before it can then think about what to sell you and ping Amazon's API and then actually sell you the thing. Um, so right now what people are working on is agents that can do that parallel processing where as you're talking it's actually taking those input tokens and, and processing in real time what it could then serve you. Um, which is very similar to web experience because if you think about it, as you browse on the web, the sidebars and all these things change dynamically in real time as well.

Speaker C: Yeah, yeah, yeah. I mean that's almost like, yeah, reminiscent of whatever, the mid-2000s when everything got asynchronous. All of a sudden you're like, oh, this is actually a lot more useful now. Um, yeah, that's great, that's interesting. Thank you.

Speaker B: Yeah, and maybe one comment too is it reminds me of ChatGPT, where ChatGPT now has ads as well. Um, Amazon's obviously always had ads. Um, but there's a slight difference where at least in a ui you have a very clear understanding of what is sponsored versus not sponsored. Um, so I'm really curious what Amazon's going to do to make it clear to the consumer what's a paid listing versus an organic listing. Because I think right now what I don't want to happen is an erosion of trust. Um, I would really like to make sure that people think of the LLM as truly trained on the public Internet, whether it's Alexa with Amazon or whether it's chatgpt. Um, and then they can have an additional like, overlay of ah, sponsored posts. But it shouldn't ideally be pay to play. Ideally it should be like truly democratized knowledge plus an overlay of pay to play if necessary.

Speaker A: Yeah, and in audio that's difficult, right? Like, what are you going to have, you're going to have Alexa kind of say it to you like, and now for the sponsored, you know, uh, product results, you know, and then that kind of ruins the experience too. And so there's like a balance, uh, there that, that needs to be struck. Um, okay, that's, that's the first bit. Uh, let's move to the next piece of news. So there was a lot of announcements from, you know, wpp, Magnite Horizon, a bunch of companies just investing more and more of these like agent to agent, uh, protocols and infrastructure and products. Um, and it's interesting to see like, who's investing in the infrastructure layer of like, hey, we're going to be the ones that enable agents, uh, to transact with each other versus like, hey, we're the ones that are, um, actually creating the agents themselves to sit on top of those protocols or that infrastructure. It's interesting to see some of the hold codes and the agencies invest in the infrastructure layer because in the past like that's not where the agencies have sat. And that's, that was kind of the, the, um, the crux of the announcements from WPP and Horizon in particular. Actually WPP was uh, the big kind of infrastructure play. So um, you know, I think it's still early. Like what I've heard from most people is agent to agent buying and you know, media buying, planning, all of that stuff is pretty nascent. Like it's more conceptual than it is reality, uh, right now. But Kashish, what are you seeing in your world? Um, and you guys, you know, you guys are kind of sitting, you're in the mix there, right? Like, so what are you seeing kind of play out?

Speaker B: Yeah, well, I mean, I'll make, I guess to address directly WP and Horizon. Um, I think it's really awesome that they are opening up the railways for people to interact agentically. Um, I think that's very forward looking and the question will be who's actually ready to interact in that way? Which brands are actually capable of this, actually ready for it, or even want this? Um, and I think I feel the same way about this concept for any agent to agent protocol that's being built today because you see these all over the place. But I still don't think they have that much adoption. My um, take, typically speaking on AI is how can we provide outcomes rather than infrastructure to potentially provide outcomes. Um, and so what I would want to probably dig into is really like what are the biggest areas where AI can help advertisers? Um, and hightouch's take on that would be for the creative and content pieces. Um, so like for the past several years we've helped advertisers do data driven advertising. It um, turns out when you have unlimited personalization, you also need more content variety, um, in order to actually leverage unlimited personalization. And so content variety is really truly one of the biggest bottlenecks we see. Um, and AI is still really bad at staying on brand. So the number one problem we hear from CMOS is I tried AI in 2025, I'm trying it again in 2026, but no matter which foundation model I use, it's not able to adhere to my brand guidelines, even when I explicitly state my brand guidelines. So it's like 60% of people don't have their brand guidelines. The 40% of people that do have brand guidelines written. Still, AI is not so good at instruction following and determinism. And so it's not adhering to those brand guidelines. And I think that's like something that the ad agencies as well as hitouch are all really well poised to solve. And I think that's a very difficult software engineering challenge. Um, it's akin to building an API is actually really easy. Um, but building an end product that produces on brand content every single time with 99.9% accuracy versus 8% accuracy. That's the very classical challenge in AI. How do you get to very strong, um, accuracy in results versus 80% results?

Speaker C: Um, Kashish, I wanted to dig into this difference between say an infrastructure layer versus activation, um, layer. You could characterize it in a number of different ways. How do you see that? Because like, yeah, agencies getting into kind of more infrastructure feels like maybe a side quest for them. There's an argument to be made that it's like, no, you guys are services. Like you kind of you, you're meant to exist at the edge. Uh, to make the best use of the infrastructure. You don't need to make the infrastructure. But like how, I mean how does hitouch see that world? How are you building forging that distinction?

Speaker B: Yeah, I think our um, so Hychin would basically make a claim that we need to help marketers do a few things right, so ideation and insights, which can be vastly more exciting when you have unlimited intelligence at your fingertips. M. So it's like we have maybe this much intelligence and we're using a very small piece for actual insight and ideation. So that's like I think the P0. Um. Another second thing once you have this ideation piece is content generation and the third piece is media buying and execution. Um, my understanding to the credit of these agencies is they want to be there for the media buying step, regardless of who is doing the media buying, whether it's an agent or a human being or something else. So I think that's really smart actually because for them, um, they don't necessarily need to be opinionated about who's creating the content. It could be a design team, it could be an AI agent or it could be themselves. But they do need to be opinionated about who's buying the media. Yeah, um, I think from that regard it's quite smart. Um, and the thing I would push agencies as well as consumers and media buyers on is how can all of this be a true feedback loop that builds intelligence over time? So maybe right now we have some ad reporting. How can we go from that simple ad reporting to actually a more AI driven feedback loop where AI is constantly seeing my ad, uh, reporting on a per content basis and a per topic basis and then developing opinions for what we can do better. And those opinions should truly be pro consumer. I think the golden state of this is always how can we use insights to actually drive a better consumer experience where people are actually seeing things that want to see.

Speaker C: Yep.

Speaker A: Yeah, I think like the, I mean it's, it's interesting because we, we talk. I'm going back to what we were like where we started here, right. Which is AI in marketing and advertising. I think at first came after the lowest hanging fruit. Right. And the lowest hanging fruit was like the 80% of planning, of even creative production of, of reporting and analytics that AI could kind of bite off and do a decent job with. And then you still need humans for like the remaining 15, 20% of those things. Especially creative. To your point.

Speaker B: Right.

Speaker A: Like it's definitely not going to get on brand even today, um, most of the time. Right. And so AI came after that stuff first. AI is now, I think media buying and execution are perfectly ripe for AI to disrupt it. But it's more complicated because there's a lot more hops, there's a lot more companies involved, there's a lot more technology involved, there's latency involved, there's a lot more commercial um, interests at play. And I think that has made AI's disruption for like media buying slower. But I actually think there's a huge upside or there's a huge upshot there in the next few years that we're going to start to see. But it's going to require a lot of this kind of like collective effort, collective work protocols, commercial kind of partnerships, things like that, to kind of get the ball moving down the field, you know.

Speaker B: Yeah, I strongly agree with that. And I'll just add when you go down to like the basics of uh, media buying, um, the reason I don't believe yet that AI should be doing media buying is because you can't yell at your AI for doing it wrong.

Speaker C: Right, Right.

Speaker B: You could turn your AI, but you also can't fire your AI. Right. So like having a media agency makes actually a lot of sense to me. For Media bank, you, um, have a throat to choke for the metric and someone that you can talk to owns the metric. So I think actually that makes a lot of sense. Um, and the question I would ask is, is there a way to continue with this kind of human to human interaction. But where there's an insights engine that the agency owns or the brand owns, either side could own it, both sides could own it together. It doesn't actually really matter. But can this insights driven engine give us opportunities that then the agency also executes on? I think that would be like the golden state that I would recommend most CMOs and like, heads of brand to think about. Um, it's actually to really, really strongly sponsor human in the loop and strongly sponsor human discretion and not to trim down your marketing teams. Um, it's to give these marketing teams significantly better insights and intelligence, um, that they can still apply their discretion to rather than just letting it run autonomously.

Speaker A: Yeah, yeah, great points.

Speaker B: Um, almost every CMO I talk to is like significantly more bought into human discretion and um, human decision making than they are into AI making decisions. But they still want AI to suggest the ideas that we would then have on.

Speaker A: Yep, yeah, do the work, suggest the ideas. You know, bring information together, be connective tissue across teams, like all of these things. But then, I mean, that's the big hypothesis, right? In the future, the highest leverage, humans are going to be great decision makers. Right. And that's true for marketing. Um, okay, let's go to the last top story. So this one wasn't like some big deal, but I thought it was interesting, which was, uh, Omnicom did a deal with Netflix to basically use AI to place virtually, uh, branded products into video content. Like, like in the moment, real time. Um, and I think we've seen iterations of this type of thing over the years, right? Of just like, oh, well, um, uh, I forget the name of the company. There's a, I think like a company called Anoki, right. They're basically, uh, looking at real time video content and then basically matchmaking and actually like changing the Coca Cola, Coca Cola bottle somebody is drinking in a Netflix show or changing the billboard that somebody sees as they drive by. You know, and so I think this is interesting as a use case specifically for AI. Um, I just thought it was interesting to, to see Omnicom partner with Netflix to do something like this. And I was wondering like where the technology was coming from too, because they didn't make that clear in the press release. It's like, okay, well, where's Omnicom? Doesn't have that tech. I'm pretty sure Netflix probably doesn't have that tech. Uh, but they didn't say who the technology layer was or, you know. So anyways, um, any quick hits on that? Any thoughts on that?

Speaker C: One that's too bad for whoever's providing that technology. Yeah, yeah, I feel bad for them now. Shiv, I, I, I was just gonna say that I don't, I mean, have, have we collectively as an industry seen evidence to date that that type of dynamic placement works? It's not a new concept. Like there's new ways of doing it now with AI, generative AI, you know, make it a lot more efficient. But, uh, I just, I've always been super skeptical of like reducing the craft of advertising down to like where everything is a spot market. And I just don't even from a media theory perspective, I'm not sure that works that you can just, I think even the act of having a break to see an ad helps the ad be more effective. When you hide the ad in the content, um, there's an argument to be made. You're actually making the ad less effective because the user's not even aware that it's a promotion. And so you're like shooting yourself in the foot. You're spending promotional budget, but it's not being couched in a, uh, promotional context. And so it'd be interesting to see if that works. But I just like, I don't know, I feel like we've been doing this for like 5, 10 years in other ways.

Speaker A: Yeah, I agree. That makes sense. Kashish, any thoughts on that or. Pass. You can say pass.

Speaker B: Really, really interesting take from Miles. I, nothing else to add there.

Speaker A: Cool. All right, rock and roll. Let's go to quick hits of the week. Um, so just a couple of things to note here. Um, actually already hit on the Amazon conversational ad, so I'm skipping that one. Zeta and Palantir did a seven year deal where Zeta, uh, is going to be basically cross selling its, uh, martech and cloud, uh, you know, infrastructure, um, solutions to Palantir customers. And Palantir is going to be, you know, providing some underlying technology for Zetas, uh, Zeta's products. And actually I would love your thoughts on this one, Kashish, because like, first, uh, of all, do you view yourself as a competitor to Zeta or are you guys complementary to what Zeta offers? And then, yeah, like what are your initial thoughts on this deal?

Speaker B: Yeah, let's understand more about it. So in Zeta we see as two things, right? We see it as a DSP of sorts where they're an ad server, um, but we also see it as a, um, email platform to the extent that it's an engagement platform. And cdp, we do compete with Zeta to the extent that they're like a dsp or you can buy ads and media from them. We're not a competitor to them. Um, and we'd actually integrate. And I'm actually really just curious, um, for your take on what the AI technology will do, because I think Palantir, with Foundry can do anything. It's the same as saying, like, you could just sort of OpenAI's models or like, I guess in this case it'd be like Claude's models from Foundry. But I'm really curious what you think that they're going to do with the actual underlying models.

Speaker A: So I, like, I started reading this and my mind started, um, going in weird directions of mostly confusion. Um, I was mostly confused when I was reading the story because I didn't quite understand, like, what the companies are providing to each other from a technology standpoint. Um, but what I did pick up on, because I'm more of a commercial person anyways, and just like, okay, well, one big part of the press release was Zeta very openly saying, like, we now get to sell our solutions to Palantir's customers, who are just not our customers today. And a lot of these customers are like, they're, they're government agencies. Right? They're. They're just like totally out of our icp. And Palantir is willing to bring us over to them now as part of this, this arrangement. And that to me was fascinating because, you know, why is Palantir interested in doing that in this arrangement, uh, and in this agreement? Like, the commercial aspect of that wasn't completely clear. Um, so anyways, uh, I was left more confused and I was hoping you would, uh, help me understand it.

Speaker B: I guess. I think I saw it as the opposite. I saw it as like, um, Zeta has these customers and they want to build some sort of ontology or AI foundation for those customers. Um, and so they're leveraging Palantir's technology in the backend to provide to their customers. And so I think the idea would be, like, if I use Zeta as my CRM and my engagement tool and my dsp, then can Palantir help me build an ontology that I can then use to basically build some intelligence on top of this data.

Speaker A: Yep.

Speaker B: Obviously that assumes that Zeta has all the data for that brand, but in theory, if Zeta has all the data, then they do need some sort of ontology on top of it and then some sort of intelligence on top of it, which in this case Palantir would provide. But I don't think I see it as, Yeah, I Don't think I see it as Palantir bringing Zeta into customers. I think of it more as Zeta has decided to leverage Palantir under the hood as a platform.

Speaker A: Yeah, I'm using AI to parse this out for me. I think you're right. I think we're both right, Kashish. Like, it's saying, you know, there's, uh, Palantir's, you know, tech in the background, will help Zeta deploy AI and agents into these large enterprises, um, you know, in a more secure way. Uh, and then the flip side of that is the commercial go to market. Palantir will actively support Zeta sales efforts by offering their marketing intelligence solutions directly to Palantir Foundry Enterprise customers. So that seems to be like, the crux, like the. The exchange, the value exchange. Um, I still don't quite know what Palantir gets out of this, other than maybe just like, a big check. Um, and I don't know what the size of that check looks like, so.

Speaker C: And you talked about them on this podcast. They got that out of it. Yeah.

Speaker A: There you go.

Speaker C: They get press release work.

Speaker A: There you go.

Speaker B: Okay.

Speaker A: That was supposed to be a quick hit of the week, and we spiraled. Uh, okay, Kashish, let's get into High Touch world, you guys. Like, we could probably just fill an entire podcast just talking about you guys. You've been all over the news in the last several months. Um, I'm just going to tell the audience some of the things you guys have been in the news for, and then let's. Let's hear your thoughts on this stuff. So, first of all, you've done a bunch of product launches. You launched Ad Studio, which, you know, I view you guys, traditionally, you were a cdp, you know, the first kind of composable cdp. And now you're going into these new areas. So you launched an AI driven creative tool called Ad Studio. Uh, you launched Lifecycle Studio, um, which I would love you to help us, uh, understand a little bit better. And then you were in the news for some really cool stuff. First of all, you did a massive fundraise. So you raised, uh, I forget the number, but it was out of $2.75 billion valuation, which is just astronomical growth, because I remember, I don't know, six months ago, you guys raised that, like, a billion. And now you're at 2.75, which is just amazing. Congrats. You're crushing it. Um, and then the other thing, the other big piece of news was you made an offer to Live Ramp, um, or sorry to Publicis who bought Live, apparently is buying Live Ramp, um, for the identity aspect of their business. Uh, and then last thing is you kind of rebranded a little bit into becoming you know, not just composable CDP or marketing automation platform, but more so an agentic data marketing, uh, platform or agentic cdp which databricks also announced kind of at the same time. So this is all the high touch news that I could find, which is a lot from the last like couple of months. Did I miss anything, Kashish?

Speaker B: I think that's spot on. That's quite a. Quite a few things. Yeah.

Speaker A: Okay. It's a lot of things. You guys have been busy. Um, so where do you want to start? Which one of those pieces of news should we start to unpack?

Speaker B: Yeah, maybe we can start with Agentix CDP and then we can go to the like lifecycle marketing and advertising studios.

Speaker A: Great, Love it.

Speaker B: I think there's fundamentally like three things that encompass most of marketing right now. Um, you have insights and data which we would classify as eugenic cdp. You have lifecycle marketing which has a vast amount of things. You have sms, email, push in app and so on. But there's all this orchestration for lifecycle marketing. And then you have advertising. And so we'll just say like these are the three things. We won't include ad sales in this. We'll just say advertising is media buying. But these are kind of like the three things. Um, and our opinion is that AI and data can help in very different ways to make these three things better. Um, and so we kind of, um, for the last several years, right, we built what we call a composable CDP. The idea was that we actually turn your existing customer 360 or customer database into a place where you can do segmentation, journey, orchestration and activation. So we gave marketers access to the data they already had. Uh, which sounds crazy, but most marketers had a bunch of data and weren't really accessing it and using it, ah, day to day for the marketing that was the last five years of our journey. And then we now have these unlimited intelligence entities that we call AI agents that can actually do work 24 7. So what we found is um, once you build this semantic understanding of your customer data and the railways to actually activate on it with audiences journeys and activation pipelines, um, most of the marketers we work with said, hey look, you're a great tool for doing what I already know I need to do. I want to do abandoned cart. You'll do abandoned cart extremely well. But what If I don't know what I need to do, what if I need ideas and suggestions and insights on which customers to target and what campaigns to run and what's working and not working, can you help me actually generate those insights autonomously? Um, and so that customer demand resulted over the last two years in us building what we call the Agentix cdp. And the Agentix CDP is inherently a proactive, um, customer data platform. The idea is you'd wake up every morning with the top five opportunities in your inbox of how you can make your marketing better, your customer experience better, or make more money. Um, and we would then allow human discretion to decide which of those opportunities we actually tackle. And the Agentix CDP would autonomously tackle those opportunities for you through the actual campaigning and the execution.

Speaker A: So it's ideation. Uh, the agents handle the ideation and kind of pulling all the information together, pose it to the human, the human says yes or no, and then the agents then continue on to execute.

Speaker B: Exactly. Spot on, yeah. Um, and it's a different paradigm because typically, um, a marketing tool is a place you go where you're like, great, I've been working on this campaign for a month. Let me execute it now. Now, it should be the place where you go to think about and ideate, like, which campaigns you think you want to run and what might work. And then you use discretion and then it helps you execute. Um, and so that brings me to the how do we help you execute? So execution has two components. It has orchestration, which is mostly software, and then it has the most important component, which is the actual content and creativity. So if you ask marketers what do they value and what do we think matters? Content is king. The best content will always win. The best technology will help on the margins, but the best content will always win. Um, and so that's why we built these two studios, which are lifecycle and advertising studios, to help marketers ideate, um, and actually build on brand content for their campaigns. Um, and so I could be any sort of like large fashion brand. Um, and maybe my Agentic CDP has decided or found out that I have this class of customers that are very high ltv and I have no content for those customers. Maybe they are people over 50 years old that have a lot of spending power, but all my advertising is for people that are 20 to 50.

Speaker C: Great.

Speaker B: I should now build a new set of content for this. That's not easy actually. That requires the AI to say, here's an opportunity. Then it requires my brand director to go hire, um, New models to take new images of and then a design and creative team to work together on actually building out this new content. So how can we insert AI into this entire workflow? It's not just like you ping Gemini, you get new content, as Miles said, that probably doesn't work. Uh, you probably will generate hallucinated content that doesn't really look on brand. So it's really about this end to end journey. Um, and then using AI to actually execute against that. I think that's where as you mentioned Shiv, it doesn't yet really matter whether AI can buy the media or not. I think everything up until that point is still like 8 to 18 weeks of work that we think could be streamlined in today's. Um, so I think a good summary would be like agentic CDP is the insights layer. Ad Studio helps you generate on brand content and run that content in production. And then lifecycle marketing helps you agentically generate your emails and your sms. Um, so it's very similar to what you might see from some of these um, email platforms, but it's really just purpose built to do the on brand email generation. What we find is everyone can do text, uh, when it comes to emails. The actual HTML and templating is really hard to get on brand. Same thing goes for ads. Image and video is really hard to get on brand. And so we've really spent the last 18 months tuning models that can make image and video and emails on brand.

Speaker C: Wow. Uh, so Kashish, like putting all of that under an umbrella, like how do you and the team conceive, how do you draw the boundaries around what High touch is going to be? Because you know, you started as a relatively niche offering, composable warehouse native cdp, great market for that, great product, great, you know, market to be in, but you've grown beyond that. And so like partly I'm asking like what's, what's your conception of like the stopping point really? Okay, this is what we're going to own because you just described a lot of stuff across a lot of different areas of market.

Speaker B: Yeah. So I would um, I would honestly just tell you the truth, which is that if it's a marketing problem, we'll probably solve it. Um, and the question is just when, um, and can we, and will we stay true to our core value, which is portability and like integration. So I think the way we view the world is marketers want an end to end platform that can help them with any of their marketing activities, but they also want best of breed. So what this means to hitouch is if the customer wants to, let's say, use Adobe Campaign or Adobe Journey Orchestration within this workflow. With Hitouch, they should 100% be able to. And we should build integration such that we work just as well with Adobe as well as we do with our own technology. Um, that way like we're not uh, forcing customers to lift and shift from an existing architecture to a new one. We're just doing the things that they don't do today and adding value. If ultimately they decide to move everything over to hitouch, that sounds great and we'll help them do that as well. But that should not be a core requirement. And I think that's what's burned marketers over the last 10 or 20 years. It's that every platform wants to do everything. You now have to go through the five year journey of actually adopting that stuff. So what if instead you only adopted the things that were valuable? So I'll give you a very clear example. Could you use hitouch for audiences and journeys and email generation? Yes, but let's say you already have an audience builder or journey builder and all you really want is agentic email generation. We can 100% provide that to you, we can integrate it for you into Iterable, Braze, um, SFMC or any email tool you have and we can just do the thing that's net new and highly valuable to you and actually doing that thing is hard. We have to ingest every one of your emails in the past, tag those, look at your CMS and your DAM and build a vector database on top of that and then we'll identify, generate your emails. So there's a lot of infrastructure work we do to do this one thing and we're very happy to just do that one thing and provide you value there. So I think our view of the world is any problem marketers give us, we're going to be very happy to solve over some timescale, but in a very portable and extensible way so that you don't have to lift and shift.

Speaker A: Hey, so Kashish, I want to like uh, so I, I totally hear you and I love that vision. I think like I would love your thoughts on. I agree with the statement that every platform wants to be everything to marketers, especially the big ones and especially today in the era of AI, I think everyone feels like A they have the permission to be everything to everybody because AI democratizes so much of this technology and then B like they can quickly speed to market is Just faster and faster and faster with like creating new products and tools using AI as you're kind of like as your, as your, you know, booster to, to get products to market, right? So um, everybody wants to be everything, whether that's Salesforce, whether that's Snowflake, whether that's Google, whether that's Agencies, right? Whether that's you guys. Like everybody wants to be everything and AI is the great accelerator of that. Now the, the issue I think, I think if you're a marketer is like not everybody, not every company or platform seems as altruistic as you guys, uh, at least in theory on like where they're willing to be interoperable, uh, and port things back and forth and then that creates, that creates issues for marketers, right? And so like, and I'm sure you've run into that, right, where you're like, hey, we're happy to be anything to everybody, but in order for us to do that we have to play nice with all of these other platforms and companies. But then the commercial people come into the meeting and they're like, sorry, we're not playing nice. Like, so where does I get, I don't even know where I'm going with this. You kind of hear what I'm saying, right? Like, how does that kind of disrupt your vision, I guess is what I'm asking, if it does at all.

Speaker B: Yeah, I think um, at the moment, like it, anything can happen, right? But I think like people vote with their feet. And so what happens is maybe um, some company will say, I'm not going to make this available as an integration. I'm not going to serve an mcp, I'm not going to serve an API. You have to use my everything closed ecosystem. But what will ultimately happen is they'll have 10 enterprise customers that won't be okay with that and those customers will demand API access. Um, and it's up to those customers who use that API, whether it's like themselves and some interface they build or whether it's hitouch. And so our ultimate belief is that even if there is some trepidation short term, that long term unopened ecosystem is necessary because a lot of these customers have a lot of budget that they command with these vendors. That's um, our belief and I hope to be correct about that belief. But I can't guarantee it.

Speaker A: Yeah, um, I mean it's a leverage game, right? It's an ongoing game of leverage between the vendors and the buyers.

Speaker B: Again, I'll give you another clear example of this. Um, I just described a pretty novel way, um, that hiturch does content gen that's on brand, yet we don't claim to or want to build the dam or the cms. We read from Adobe and we write to Adobe, um, or any dam we choose. Contentful doesn't really matter. Like any dam we'll just read and write from, which is actually exactly what marketers want. They have zero desire to go and do a lift and shift of the dam they spent 10 years building. What they want is for hitouch to add tags to the dam they already have and to upload Net new content to the dam they already have. So I think it's just natural that this kind of interface will continue to get better and better. Um, and then it's also possible that some companies will go for closed ecosystem and then those companies will become losers over time in terms of customer demand.

Speaker A: I just, I wonder, I mean my brain always goes towards paid ads and you know, it's the, the Amazons and the Googles and the metas of the world like will probably never play nice. And yeah, you know, they hide behind the data privacy thing a little bit. Um, but they're always going to be sitting there as like, hey, they have a ton of leverage because they have a lot of really, really, really valuable consumer data. Um, but if they're not going to play nice, like who's, who's willing to kind of say I'm okay, you know, I'm okay not using meta for something or Google for something. Right. They're the 300 pound gorilla that everyone's got to deal with and overcome in some shape or form.

Speaker B: That's a great point. If you own the eyeball, I think you have a lot of leverage. And so I think even ChatGPT and Anthropic soon, so have that same leverage over advertisers. And then obviously the trade desk is much more open and wants to give back that telemetry and data to the brands.

Speaker A: Yep.

Speaker B: So I think like the best we can hope for is really like if it's first party data, brand should own it and then be able to do whatever they want with it. But yeah, I don't. The third party data is changing.

Speaker C: Yep.

Speaker A: Um, okay, tell us about your live ramp play. Well, let's close on that. So what's the thinking? Why'd you guys make uh, that offer? Like, is there any traction? Are we going to break some news here on the podcast? Like, are you getting it from Pubisys? Does Pubisys have cold Feet on the deal. What's going on over here?

Speaker B: Yeah, so I can't comment on the actual deal itself, um, or anything from Pubisys, um, but I can talk about like our intention behind it.

Speaker A: Okay.

Speaker B: Um, Liveramp has a data asset and our desire is to buy just the data asset piece of the company.

Speaker A: And can you, can you explain the specific asset a little bit more in detail?

Speaker B: They would be their identity spine. Um, and the idea would be like, they have this identity spine that they've sold to many, many different brands. And could we just adopt that identity spine and make it part of hitouch? Um, it's typically an identity spine that was based on third party cookies and um, ramp id. Um, and so the idea would be for hitouch to own that identity spine, um, for the purpose of making it accessible and giving it to the Hydra's customer base. So I think like right now it's um, like you kind of have two businesses to leg ramp, right? You have the identity business and then you have the rest of their business. Um, and then you have all the employees. We'd still want publicists to keep the rest and the employees, uh, which would include the data collaboration and data clean room business and their cross media intelligence, which is, um, like also for data collaboration. But I think like live Ramp ID would be really kind of up in the air. It's something that we would have to figure out because I think publicists would in theory want that and we would in theory want that as well.

Speaker C: Yeah.

Speaker A: Uh, I mean, I. Go ahead, Miles.

Speaker C: Well, Kashish, can you expand a bit on like, I can see how the ecosystem as a whole might benefit from high, uh, touch taking that identity spine and keeping it, uh, you know, agnostic neutral. What's in it for publicists to sort of separate that? Like what, what, how, how do you think about, like. So you, you, you know, characterize sort of like there's the identity spine and there's everything else. Right. Uh, what you're saying is it would imply that the, everything else has a ton of value for publicists. And that's the part that I don't. I literally, I'm asking this out of ignorance, not that I've got some like, fully formed opinion. What's the value of all that other stuff?

Speaker B: Um, I think the data collaboration actually is extremely valuable. Um, it's not really part of our core business, but I think where it would show up particularly is data sharing, um, between brands and advertisers, or, sorry, advertisers and um, ad Networks.

Speaker C: Okay.

Speaker B: Um, a lot of these agencies, for example, are actually running media networks and retail media networks. And marrying the data between the ad sales and the media buying becomes really interesting because then you get real intent, um, as well as all of the telemetry and integrations. So we are unique, right, because we've already built most of the integrations that Liveramp has built. Um, those could be really helpful to a company like Koopsys. Um, we also have an identity spine, and we compete pretty regularly with Liveramp. So for us, it's not really about getting the identity spine either. Um, it's more just that they have this set of very, very like a very large set of customers that are maybe not publicist customers. And so we'd want to help those customers stay neutral, um, and be able to continue having choice and not really be told in one direction or another.

Speaker C: Yeah.

Speaker A: Cool. Um, I have so many more questions about this potential deal. Um, but we don't have time for all of them. So I guess, uh, my last question for you is. Tell us about the databricks thing. We'll end on the databricks thing because I thought that was fascinating. I think, like, the announcements happened within the same week where you guys said, hey, we're in a Gentix cdp, and databricks said the same thing. They're coming at it from a different place. But are you guys kind of meeting in the middle? Like, that's fascinating to me.

Speaker C: Yeah.

Speaker B: So, I mean, um, we've known about databricks building this for about a year. Um, we knew and I think, like net net, we're pretty excited. So let me, Let me give you, like, kind of like how we think about it, because I think the market thinks about it differently. Uh, how we think about it is high touch is only usable if you build a strong customer. 360 in a place like Databricks, Snowflake, Synapse, and so on. And we've always wanted for databricks to build things like identity resolution, um, and ingest capabilities and just things that make it easier to onboard and get customer data in there. Yeah, um, we think databricks is a great option for that. Um, so are Snowflake and Bigquery. Um, and so we think like, the warehouse customer Data lake is where data lives and should continue to live. Um, but that high touch is where marketing gets done. Um, and most people would probably either agree or be able to agree with the fact that marketers don't log into databricks today. Um, and we don't foresee that they will change and start logging into databricks tomorrow. We foresee that marketers will want a purpose built tool, um, where they can do kind of their end to end workflow. Um, and that that purpose built tool should integrate really closely with databricks. Um, so I think databricks customer lake is a massive threat to CDPs that store data.

Speaker C: Right.

Speaker B: Because those CDPs should not have existed in the first place. And we have been spending the last five years trying to put those CDPs out of business. Um, this just helps further in doing that because now you can even furthermore build your actual customer360 and databricks and not use a CDP that stores data. Um, I think Adobe and Salesforce for example would like to be that data lake. And now we have even less reason to use them as that data lake but as the marketing layer. The marketing layer still is Salesforce or Adobe or Hightouch and these are the three that kind of compete with each other. And then databricks are the data foundation that none of us really compete with in my opinion. Um, and I think that's how it's going to pan out. So even if databricks does build their audience and capability which they announced a couple weeks ago, um, and some of their other capabilities, net net, there's still a lot of marketing surface area left that they're not going to build. Um, and I think we'll be up for grabs especially as an AI native solution starts existing which would be hitouch. So that's our perspective and um, we're continuing to partner with databricks so we have always partnered with databricks. We're keep doing that. Um, there might be confusion in the field. I think that really is unfortunate where like their reps and our reps are confused about which one to sell. But our job provide clarity on hey look like we still co sign on using databricks and we want High Touch to be the layer on top of that that makes marketers use um, the thing.

Speaker C: Yeah.

Speaker B: Um, and then having the agentic marketing platform with the content generation, the execution, um, and the channels will be just another reason for databricks reps to know when to bring in High Touch because they know that that layer is another tool outside of databricks.

Speaker A: Makes sense. No smoke. I was hoping for more smoke.

Speaker B: Oh well no, we, it's just shut it down. Yeah, we, I mean we respect those guys immensely.

Speaker A: Yeah. Uh, awesome. Hey Kashish, this was awesome. Thanks for being on the pod. Really appreciate you, uh, you know, keep it up. You guys are, again, just rocket ship, crushing it.

Speaker B: Thanks.

Speaker A: Thank you for sponsoring our last AI accelerator. Uh, I think you guys presented Ad Studio in one of the sessions, and people were thrilled about it. They were really leaned in and thought it was a really cool product. I personally think it's so cool to kind of have an ad creation product sitting so closely with the customer's data, with the marketer's data. It just makes tons of sense. And so kudos to you guys for always pushing the envelope and being innovative. Um, thanks again. Thanks for being on the pod.

Speaker B: No, thanks for having us. And M. Yeah, just looking forward to rest of 2026.

Speaker A: Yep. All right. Thanks, Miles. Thanks to our listeners. We'll see you guys again in a couple of weeks.

Speaker B: Thank you.