Name: Martech for 2026 Keynote
Uploaded: 2025-12-02T17:17:07.429Z
Duration: 1 h 1 min 8 s
Description: Martech for 2026 Keynote

Transcript for "Martech for 2026 Keynote": Good morning. Good afternoon. Good evening. Yeah. Good night. Alright. If that isn't too much of a Truman Show thing here. Welcome to Martech for 2026. We're so happy you joined us. I'm Scott Brinker of, chief Martech. And I'm From Martech Tribe. So great you tuned in again. It is wonderful to be here with you, my friend. So today, we're gonna go through some of the latest findings, of our new MarTech for 2026 report, but we don't wanna keep you in suspense. I know, you came here for the report as well as the witty repartee between me and Franz. But you can get the report right away. If you go to that docs tab, in the right bar here on Goldcast, you should be able to download the copy right away. And then in the meantime, we will give you a walk through of some of our favorite findings for this next hour. After that, Franz and I are gonna have a series of candid interviews with experts and executives from each of our seven sponsors, Growth Loop, Hightouch, Intuit Mailchimp, MetaRouter, Progress, SAS, and Treasure Data. Just enormous gratitude, to these sponsors who made it possible for us to spend these past six months working on this research, producing it, and now being able to give it to you at absolutely no cost. So we hope you'll stick around for these discussions with each of them. These aren't sales pitches. These are gonna be some really cool deep dives of how they see the evolution of Martech coming up here in 2026. So let's face it. Things are just moving at a crazy pace. You know, I'm old enough to remember once upon a time, there was this thing just called time, and it moved at a pretty steady pace. And then the Internet came along, and everybody started talking about Internet time, which was seven times faster than normal time. We measured it in dog years. And then, on these past few years, now people are talking about AI speed, which is like some multiple on top of Internet, time. So, yeah, if if you're feeling a little dizzy, I mean, I think I am too. But, whoo, Franz and I, we've been doing our best here for these past four years to really, like, make sure that we're able to keep you up to speed with all of these changes. Yeah. 100%. And it's all about perception time, of course. Time is the same, but, yeah, I heard this nice, saying of you can, not add days to your life, but life to your days. And it feels more dense and condensed. All the stuff that is happening on a weekly, daily basis almost is so hard to keep up. But, this is exactly why we're doing all this research to figure out, you know, what's what's really happening? What's what's the thing, this narrative, the story line here? And it really helps us to make sense of, you know, what what is the market and where is it headed. And we just briefly look back so we can give you a theme for today and for the report that we are sharing with you today in the docs section. So looking back, we started in 2022 with the reports and the webinars and, of course, it started with the MarTech Map launch, the interactive version of the MarTech Map. But we also came up with concepts and models like maturity model for MarTech and stacks, aggregation, also the big ops, optimization concepts that we unveiled. And then also, you know, we went into the Jenny I era more and more. We found how the long tail works, how it's disrupting maybe, the head and the and the torso or not, how stacks are being rationalized on an ongoing basis. Then we did a deep dive and deserved some extra attention, which is composability, market architecture with the data layer and how stacks are developing value differently across industries. Then about Gen AI use cases and strategies, brand LLM, and HyperTale. Oh my word. There's a lot of groundwork covered, Scott, so far. The hurdle moment in Martech. No. Absolutely not. And then well, just like I said, we're trying to figure out, okay, what's this theme for now? And, we'll show you in a second, unpack the theme, but the theme is basically about AI agents are augmenting MarTech. There's more to that than just a title. What we see in the market in on LinkedIn, of course, is a lot of, SaaS is dead type of post. AI is replacing it. A lot of tombstones over here, the dates on the tombstones for SaaS dying differ, so the experts don't agree. And we certainly don't agree with the fact that SaaS is dead. It's not for the sake of not declaring a debt, but the numbers that we have and the survey results that we found are just not showing that. It's actually not so much, AI replacing SaaS. It's AI is augmenting SaaS and software and martech. So, yeah, we just, you know, stole this famous phrase of AI is not coming for your job at the person using AI. Okay. Here we go. AI won't replace SaaS with SaaS using AI will. So where does that leave us? Yeah. Well, you know, the depth of SaaS is, greatly exaggerated, as they say. I think it's mostly because, the LinkedIn algorithm must reward, you if you do a post that says blank blank is dead. You know? Why else would people like us keep doing this? They've been declaring email as dead for, like, you know, every other month. I mean, you know, emails died more often than, like, Michael Myers and these Halloween movies. But alright. Anyways, but we actually we brought data. We brought the receipts about what what is actually happening with AI and SaaS. Exactly. And we asked the question, so are AI agents replacing or augmenting your MarTech SaaS apps? Yes. 30% said, yes. It's replacing functionality use cases. And, let's be fair. SaaS can't be perfect in every corner of MarTech. So it's it's logical that people say, yeah, this is where AI does a better job. But if we then look at net new functionality, net new use cases that have been implemented or, enhancing existing capabilities, then, oh my god, you see percentages that are way higher. So these are things we could not do before and what we call net new AI capabilities. So AI has been augmenting MarTech. And what is very important is that we start to look maybe differently at how we perceive the changes, the heights, how, AI is changing our world. It's not dominating it, taking it over. It is enhancing it in a nuanced way. And even if we unpack this very graph, then look what happens if you split it into SMB and enterprise. There again, you see a difference. You see that, SMB is a slightly, more, or is leaning more towards entirely new functionality, whereas enterprise is more replacing function functionality. Yeah. I mean, it just kinda makes sense to me in the sense that, you know, enterprises typically have very, very large tech stacks, but unusual for it to be even hundreds of products in their martech portfolio. And so they probably have a lot more use cases in which they're like, oh, actually, maybe we could use this AI product for this instead. SMBs tend to be a little bit more lean, and anything they're using, they're probably using it because it is actually a necessary part of their business. But I found it interesting that, yeah, SMBs are also more willing to experiment, like, twice as likely to experiment with entirely new functionality with AI compared to their enterprise, brethren. And, you know, I mean, again, this sort of fits into the, you know, common trope, that SMBs just they're a little bit more agile. They're a little bit more hungry. Maybe in some cases, they have a little less compliance governance craziness to, like, navigate for better and worse. You know? But seeing that play out here with AI, super fascinating. Exactly. And and this is how you will see us unpacking this webinar, how we look at the market, how we look at our data. SaaS is not that. It's augmented by AI, and then it's done in a different way. And then, if you look at SMB and enterprise level, even there, you have more nuance to it. So that is how we like to look at it and say, maybe we should grow as a community a little bit our tech maturity. And how do we do that? Well, maybe we can move from binary choices to managing the trade offs. This is what outperformers do. Mature companies, outperformers in industries do this. So they don't go like, let's say AI is eating SaaS. It's more like, how can we leverage and combine SaaS and AI where possible? So it's less of, okay, yes or no, and it's more of when this happens, then we use more AI or more SaaS. And, yeah, there's a very big difference between the two. SaaS is deterministic, AI is probabilistic. They're not mutually exclusive. So you need both. You can use both. You just need to know where and when and in which situations. And if you do that, then suddenly your hype cycle turn into a learning curve, and I think that is benefiting your career, your company, the market. Awesome. Alright. Well, with that as, our intro, we are going to dive in to three themes, from the report. We're gonna talk about embracing three different domains of AI agents and marketing. We'll then get a little bit more techy, and talk about how you actually compose hybrid stacks of SAS and AI, to be able to serve these agent domains. And then we'll, close out with, frankly, my favorite topic, which is all about you, and the work that you are doing and can do into 2026, to be the most important agent of all, the change agent in helping your company navigate, this new marketing and martech future. So let's dive right in. We'll start with these three different agent domains. Now, you know, Franz and I, we'd love to categorize things. This is how we make sense of the world, you know, the chaos of all these things flying at us, you know, a million miles an hour. And so this year, 2025, has certainly been the year of agents. You can't wake up in the morning without being assaulted with all sorts of agent this and agent that. So what we tried to do was, like, look at, like, three very high level buckets of the kinds of AI agents that marketers are working with. And the three buckets we put them in is there's agents for marketers. These are agents that marketers use behind the scenes to, like, do brainstorming, production, research, analysis, distribution, all of these marketing activities that we can accelerate with AI, certainly from an efficiency perspective. But also in some of the most exciting use cases, we're also using this to bring new kinds of creativity and new possibilities to what marketers can create to drive the effectiveness as well. There's another set of agents that, we control as marketers, or as the business more broadly, that we deploy to customers, things like customer service agents, you know, on the website, AI SDRs, love them or hate them, shopper concierges, AI shopper concierges on, like, ecommerce stores. And these are things that marketers are the ones who control these agents, but they interface and interact with our prospects and customers. But the most interesting domain to us is this third category, agents of customers. These are AI agents that consumers, prospects, customers, they control, things like ChatGPP or Cloud, you know, and they can use these to shape their journey of how they interact with our businesses. We as marketers, we don't control these. In many cases, we don't even have a lot of great visibility into what's happening with them, but it's becoming increasingly important for us to be able to influence that. And so when we looked at, you know, the different types of agents or MarTech support for agents that, people have been deploying this year, Again, probably not a surprise, most of the AI agents that have been adopted in marketing are these agents for marketers, tools that we're able to use behind the scenes to accelerate and amplify what marketing's doing. Almost everyone has at least one, you know, customer facing AI agent that they've deployed at this point in time. Most of them, again, mean these customer service agents. You know, we've had these on websites for years, but let's be honest. For years, they've mostly sucked. But this past year or year and a half, I've gotten pretty good because using these LLM engines, they're just much better at conversational interaction. And, also, we're getting better at hooking them up to the right kind of data to provide more intelligent interaction with customers. We'll talk about that a bit further on. And then when it comes to agents of customers, okay, well, we don't actually control those or deploy those. What we can do is, as marketers, we can deploy technologies and tactics to help influence and engage with those agents, and we'll talk about them too. Now, you know, Franz and I just love a nice logo landscape, slide. So, you know, we started putting just a a a tiny fraction, you know, of some of these AI, agents onto a map just to give you a sense of, you know, what what is happening in these different categories. There's actually a more expanded version of this one that should be in your docs, tab there if you wanna download it. A special thing just for you being here at this webinar. So let's talk about how these things are actually being deployed. If we looked at the deployment of agents or agent supporting capabilities, across the results from our survey, I just wanna focus for a moment on the top three. The number one use case is content production agents. These are agents for marketers, you know, going all the way back to the launch of Chat GPT and, like, hey. Help me write this blog post. You know, marketers have gotten better and better at leveraging these AI tools to accelerate, augment, and amplify what they're creating. It's not just help me write a blog post now. Right? It's, things for, you know, multimedia, you know, incredible, like, video editing, now even increasingly video generation, translation, amazing tools there. Second most popular use case isn't an AI agent of of our own, but it's about, a, approach to a street strategically deploying AI optimized content to be able to influence those agents of customers, particularly TapGPP and Gemini and Cloud and, you know, all these places where perplexity, you know, where people are looking to find us is the new AI search. And then the third one is, yeah, at least customer service, you know, agents. That, again, they've become pretty ubiquitous. So we have, like, one of each kind in the top three. You can read more of the details in the report, of the distribution of the others. Just zooming in a little bit on this, you know, aside from content production agents, we also are seeing marketers, you know, about 40% or so using these more for essentially data analysis, in particular audience discovery or segmentation. 35% were using it for, like, oh, let's research competitors or let's research particular prospects. When we look at we we broke out the data for a number of these things between, like, b to b, versus b to c and, you know, joint b to c, b to b, companies. And it's interesting to see some significant differences. You know, b two c, tends to have a lot more variety of content production, that they need to do. So that's why you seem a little bit larger there. But it is content production that is by far the most popular use case, like, you know, it's a good for b to b 22 points ahead or b to c, like, 30 points ahead, you know, of that, like, data discovery use case for audience discovery or segmentation, and then goes down from there. You know, there's a little bit of differences in, like, the kinds of agents that b to b versus b to c deploy. You know, b to b much more as these sort of, like, sales assistant agents, you know, helping reps, you know, while customer journey builders tend to be more popular in b to c. So you can look at some of the different examples there. When we look at, agents being deployed for customers, again, customer service chatbots, by far the most popular. It's a pretty significant drop, down to the next use cases of, oh, using it for outbound email outreach. This is a little bit of that AI, BDR, SDR use case. Curious how many of you feel like you're receiving too many emails from AI, SDRs, and BDRs. How do you feel about that? You know, we've we've been noting the comment. You know, customer service messaging agents, inbound message response, you know, interesting use cases, but the, you know, again, the popularity of the deployment here in 2025, you know, drops under 20% for almost all of those. We'll see where they head here in 2026. Again, a little bit of a difference in the kinds of use cases that people are doing in b two b versus b two c. I mean, even if we look at just that first one, the customer service agent chatbot, much more popular in B2C use cases, less so in B2B. Our hypothesis of this is, again, the customer really only wants to deal with one of these agents if it's actually able to resolve something for them more quickly. And one of the great things about b two c is while the volume is much larger, typically, the nature of what people are purchasing as a product or service is a little bit simpler. So being able to get a much higher percentage of those issues resolved through an AI agent is why we believe that's more popular there. B two b, it tends to get more complex. It tends to get more relationship oriented where you actually have a customer success manager you might call or something like that. So that's why we think it's a little less popular there at the moment. Now, Roger and I talk a lot about, this, you know, like, when you when you're applying really any technology to customer experience, any sort of automation, it helps to make sure you're keeping in mind, are you doing things that drive greater efficiency for you as the marketer, as your company? Or are you deploying things that make life better and more delightful and more efficient for your customers? We've talked about this in the past, and we absolutely see this is one of these models you wanna bring to the agents for customer deployment. You know? Is it's all too easy to say, like, oh, yeah. We'll just go off and have these agents, and they'll contact all these people, and they'll do all this work for me. And, man, my life has just become so efficient. You know? But if the experience for the recipient on the other end, the actual customer or prospect, is not making their life more efficient or more delightful, let's just say that's not gonna end well for you. So, with great power and response great power comes great responsibility. I I knew there was a Spiderman quote in here somewhere. The most exciting one, though, for us is this category of agents of customers. And you've heard a lot of these stats here over this year. We've got some of the most recent ones from McKinsey. 50% of consumers said they're already using AI powered search. This is resulted because people are having more of these engagements within the AI agent instead of just clicking a link and going to a website. Somewhere between 20 to 50% of the direct traffic to websites, you know, is now down or expected to go down. And then even with things like ChatGPP announcing, instant checkout, Google is now letting you do things with, Google payments here as well too. I mean, we're getting to a place where these AI agents are gonna start to actually be able to handle the transactions without even people coming necessarily to our website. And McKinsey had an estimate by 2028. That might be $750,000,000,000 that consumers spend that shifts into that channel. So this is this is clearly well, all all the fun stuff happening on agents for marketers and agents for customers is exciting. In many ways, it's very incremental of just making better what we do today versus these agents with customers, we see it's really disruptive. You know? A lot of ways, like, we know you recognize this is the explosion of this category of AEO, AI search optimization versus, you know, classic SEO, because even though we don't control those agents of customers, boy, we recognize how critical it is to influence them. And so, again, like the data show, like, you largely acknowledge this. You're publishing AI optimized content to achieve that. And then, you know, some of these other tactics for how might we start to engage with these machines operating on our customers' behalf, still pretty early. One thing that sort of stood out to us is, well, you know, 63% of the respondents, are will claim, like, hey. We are actually publishing AI optimized content to influence influence those agents. Only, like, 13.6% were using some sort of technology to actually measure, are they being included within these different AI engines? Are they keeping track of the referrals, and the links that do come to their site from these engines? It's a pretty big gap. Like, if you measure like, if this is really this important to you, it might be might be good to measure it. And, actually, we suspect here in 2026, this will be something that the this gap will close pretty quickly. You know, a few minor differences here in b to b versus, b to c. And, you know, probably the main one being, like, okay. B to b has clearly jumped, on the bandwagon of, like, okay. We need to really make sure we're able to, you know, produce more and more of our thought leadership content, you know, and make sure it gets included by AI. On the other hand, b two b is also, like, less likely to be actually measuring, that inclusion as well too. Alright. My b two b friends, the download is slowing down. You've gotta close that gap here over the next year. Alright. So just barely scratched the surface. A lot more in the report there about those two kinds of agents, but how about we shift to talk a little bit about how you actually assemble these things. You know, in our State of MarTech 2025 report back from May, we had put forth this, you know, pretty conceptual diagram. But just trying to, like, make the point that it's exactly as you were saying at the beginning of this, Franz, you know, AI isn't replacing SaaS. You know? SaaS with AI is replacing SaaS. In almost all of the, marketers and organizations we talk to, the core market stack is alive and well. You know? Now it's evolving, but what we generally see is, you know, this convergence around a really strong universal data layer, you know, data clouds, cloud data warehouses, things like that. On top of that, we are still increasingly using things like, CDPs or DAMs or CRMs to provide a way of, actually doing governance and packaging of this data and content to make it more accessible to applications above that. We started to see some that next layer is, you know, inclusion of things like, you know, custom, LLM or ML, AI models. The use of, AI decisioning tools is now starting to rise as well too. And on top of that, you've got a lot of the systems that we're still using for the actual engagement with customers, you know, marketing automation or ESPs or, you know, DXPs. That's all. It's like an alphabet soup of, acronyms. And then it's around that where we're seeing most of these AI agents and tools to help with, like, AI orchestration. Even gave a little shout out to that yellow box at the very top of, like, oh, you know, it's those AI agents being run by customers. They're not technically part of our tech stack, but part of the way we're gonna have to interact with these things moving forward is actually at a tech stack level. So alright. More to come on that as well too. Franz, you wanna talk a little bit about how people are actually integrating these things? Yes. How how to glue everything together? SaaS, AI, probabilistic, deterministic. How how are you guys doing and doing that? Well, here you go. This is some of the results that we want to share with you. How are AI agents tools, agentic workflows integrated in the tech stack? Because they are some way, shape, or form married, together in every company stack. And, we see that custom built integrations are relatively high, 56% prebuilt, almost at 50, and then 40% is iPaaS iPaaS integrations, the the likes of Make and Zapier, but also NAden, Verkado. So that's a a different way of, integrating and making sure it's not mutually exclusive, so people using this, next to each other. But if you look at and unpack it at an SMB and enterprise level, you do see bigger differences. So iPaaS is really, for SMB, something they play with, try out, glue their, what is it, their hacks or stacks together. And then where you see enterprise, of course, makes a lot of sense. Custom built integrations, they have integration teams, system integration teams, they have IT, that have to do with compliance and governance and security. So it makes a lot of sense that they want to have custom built integrations of their AI into their current, MarTech stack. Now if you ask what are we integrating, of course, the systems that, Scott showed in the MarTech architecture are not isolated. They are integrated because the cost customer data has to flow. And one of the major sources for customer data is, of course, the cloud data warehouse, data warehouse, data lakes, what have you. So almost 40% has already integrated their AI agents directly into the data clouds. That is a pretty big number, Scott. Yeah. Well and, you know, it's probably even larger than that in the sense of this question where we're just trying to get folks who are directly integrating with a cloud data warehouse because folks like Databricks and Snowflake, you know, it made it possible to actually run AI agents natively in their environment. But when you think about, like, since all AI agents feed on data, there's also a lot of the cases where the data might have flowed through at one point in time the data cloud, but then it was, absorbed and packaged up into a higher level governance application like a CDP or something like that. And then that's the thing that AI agents are interacting with. It's gonna be really interesting over this next year to see, you know, where that balance is of how many things directly work with the warehouse and how many things work directly with some sort of, you know, more governed intermediaries. Yes. Definitely. And and customer data and the cloud data, lakes warehouses, it's not the only thing you can integrate with. There are more internal sources you can integrate with. And what is really standing out, at least to us, is that on top, you find this in the 60, of course, CRM, CDP profiles that we integrate with, but also brand and marketing assets. And that makes a lot of sense because number one is like, who is a customer? Let's look at our CRM, CDP profiles. And what should we tell them and how should we talk back and generate, assets? That's the second one. So, kind of surprising and not at the same time. Great to see that. Even brand voice is number four style guide. And the emails, of course, talking back to, to customers. It's it's a it's a very nice insight here. That's one of the things that stood out to me is it's not just the structured data of things like CRMs and CVPs, but it is that unstructured data, you know, of these brand assets like, emails, knowledge bases. You know, we were talking earlier about, like, how, AI is being used to enhance, or create new use cases that just weren't possible with the core, deterministic SaaS. Working with unstructured data like this, I think, is clearly one of the best examples of these are whole new, use cases unlocked. Definitely. And, again, internal data is not the only way you can, integrate your stack with, AI agents. It's also external sources. That is a wider variety. Somehow we get more creative. We have more options here. It's not one standing out, but a whole bunch. So it's prospecting. It's going to find out, is there any data or content that we can find around the customer profiles and the prospect? Can we enrich it? Can we look at maybe competition? What are they doing? How far are they? Can we find intent data, or can we maybe enrich with third party data websites our own information, or maybe directly communicate with them somehow through social media or just scrape the social media and get an idea what the most important trends are. So there's more ways of integration. There's MCP. And, yes, I think it started to grow this year, started end of last year, short period of time. MCP is something that, Claude came up with and Tropic, as a standard. Of course, it exists longer than that, but, it's not only when you invent it, but also when it's popularized. And what we see is it's mainly used in AI assist assistance so far, not yet so much in iPaaS or in core marketing platforms. And that was something that really stood out here. So, yeah, we're looking forward to where MCPs will, take over and how they will develop in the near future. And then if we unpack that, then we do see, again, in SMB and enterprise, big, big differences, especially in the iPaaS, section. We saw this earlier where, SMB is really utilizing iPaaS to the fullest. Well, here again, SMBs prefer iPaaS, enterprise prefer more AI assistance, but also embedded into their existing platforms. Very, very nice nuance to see here. Cool stuff. Well, I'm very interested to see how MCP continues to evolve. I keep keep getting announcements, it feels like every week, of more MarTech products that are releasing MCP servers and MCP client capabilities. So that's just becoming easier and easier to incorporate some of these AI capabilities even within your core MarTech marketing automation products. And so, this is maybe a good point to, like, step back and talk a little bit about this balance between, you know, original deterministic SaaS, you know, these very, like, rule based automations that, we've come to love, over the past ten, fifteen years of marketing, you know, deterministic workflows, sequences, things that just follow this this preprogram logic of if this, then that, versus, like, on this other end of the spectrum when we talk about agents. Boy, some of the vision that's held out of what the future will be is like, oh, well, there'll be this one agent It'll come up with a high level like, we'll have a high level goal we give it. It will come up with a plan, and we'll organize a network of all these other agents to go and execute that. I will be sitting back on a beach with a pina colada. You know? And so there's sometimes this, like, tension that, you know, high portion binary choice of the hype curve is like, oh, well, it's either, you know, that old school, you know, deterministic automation, or it's this new school agents do it all. The reality is it's actually a blend, between these. You know? We see, you know, we just sort of, like, sketched a few examples along this continuum, you know, where we're now seeing, like, even in environments where people are large largely running deterministic workflows, deterministic sequences, they'll, like, include little pieces of these more probabilistic AI capabilities. You know? Like, oh, I actually want to, like, interpret this incoming email and be able to, like, flag what the theme is, what the priority is, or where I should redirect it, you know. But then the rest of the workflow proceeds in a very deterministic and predictable way. You know, you can get more advanced of, like, oh, well, maybe this AI, it can, like, you know, based on a customer's website, make a decision about what sort of segment to place them in, and then decide which path to take deterministically. Maybe it's creating content, you know, at the end of that path where it uses an LLM to synthesize a a a a true one to one type message. Things like you start getting into AI decisioning that particularly if it has some sort of feedback loop where the AI is, deploying things out to prospects and customers, and then based on those reactions, is updating its model of how to improve, its performance over time. And so this whole spectrum of possibilities is how you see, you know, again, instead of the binary choice, it is the trade offs. And what are we trading off here? Which is a little bit about this sort of blue green section here at the bottom, and we go into a lot more detail on this in the report, but this isn't a maturity model. Going from left to right does not necessarily make you better. It's just these things do very different kinds of capabilities. You know, on the left, things are very controlled. They're repeatable. They're predictable. You know, like, if you feed, you know, this 100 inputs into the if this, then else, you know, deterministic workflow, you're always if it's the same 100 inputs, you're always gonna get the same 100, outputs on the other end. Versus, like, when you get into the more nondeterministic and probabilistic sort of, generative AI engines, Well, in some ways, they're, you know, actually more adaptable. They can, you know, adapt to new kinds of use cases or new kinds of data, without breaking. There are, in many ways, a lot more robust. But they have higher variance, you know, and you can't always perfectly explain why it made a particular decision the way it did. You know, there there might be minor variations in the output. In some cases, there might be major variations in the output. And, again, this isn't necessarily bad if you're harnessing it in the right sorts of use cases. It does things with deterministic, you know, sort of automation. Well, it's almost impossible to do, but they're different. And so really, like, being able to balance these trade offs is what we think is the right approach for 2026. But what other challenges are we wrestling with, my friend? Yeah. I think that everybody wishes it for one simple thing to be. The opposite is true. It's a blend of complex things. It's it's data quality, our missing data. It's organization, you know, the the process readiness, the skills gaps, maybe even ownership that's lacking. And then, did we cover integration? Yes. We did. So integration frictions are also big part of, the top three that all have more than 50% of the respondents. So it's like this is a challenge. So this is definitely something that we will see many companies struggle with and covering. And again, here we break it out into SMB and enterprise. You see that especially enterprise, I wouldn't say struggling, but has the wider variety of challenges than SMBs. And it has to do with security, of course, with governance, with cost observability, but also with, you know, all those silos in the organization and making sure people are, I don't know, reskilled, to to cross the skills gap and integration friction. So I think, yeah, enterprise has their work cut out for them, but I think they also have the amount of resources to do so. Now another breakdown could be b two b and b two c. And here we see that b two c, also slash b two b and b two c together, have a bit more of a challenge, but not the same challenges as enterprise necessarily, but it's more the lack of unified customer IDs, also governance, maybe unreliable LLM output, like the probabilistic stuff that we've been talking about. So, yeah, we're trying to figure out how how that works. So integration friction is also something you see with b to b, and b to c, as a bigger gap. And unreliable LM output is more something that b to b is struggling with because they have to really be precise in how they output. Now if we look at, how are companies using AI agents embedded in existing MarTech, or are they using maybe tools to configure, it configure AI agents with low code, no code, iPaaS, and what have you, then a large portion, over 60% is using AI or agentic capabilities inside their platforms, the existing ones like marketing automation, CRM, CDP. Like we said, it's enhancing, SaaS. And the SaaS is using AI oh, there we go. Yeah. It's not AI, that's replacing your assets. That's with AI replacing oh, sorry. There you go. That's what's happening here. Right? Yeah. The numbers are clear. And, yeah, if you look at the configuration of low code, no code tools, that's also a pretty high number, and, it was really standing out. And, yeah, I think, Scott, this is coming back to the Hyper Channel mainly, if I'm not mistaken. Yeah. This is something we, you know, we have introduced here over this past year is, you know, obviously, for the past, yikes, fourteen, fifteen years of charting the MarTech landscape. We've talked a lot about the long tail, you know, like this just expansion of thousands and thousands of highly specialized commercially packaged solutions available. But what had really started to grow over this past couple years, and it's accelerating like mad, is not just the purchasing of commercial martech software, but increasingly, companies building some of their own software as well too. You know, AI is accelerating this. You know, I mean, this is even for, like, professional software developers in, like, IT, who are using more AI assisted coding, environments to accelerate the work they do. You start to get into these citizen developed apps, the classic no code apps that, now we start talking about as being live coding. We'll talk about that in a sec. But then also these agents, you know, things like TapGPT and Cloud that can, like, actually create a little program on the fly in response to a, you know, task you give it, execute that program, and just deliver the results without you necessarily even knowing that software was created or executed on your behalf. But, you know, if there's one version of this that's probably gotten the most attention this year, it's very much the, vibe coding. You know, when people talk about vibe coding, one of the most popular, platforms is Lovable. We had a tiny example we gave of, like, oh, just going to Lovable and describing in natural language. I'd like this little, you know, calculator for customer service cost optimization that I can put on my website. And literally, in a matter of, like, a couple of minutes, it's able to generate it. Now this has limits. And so what you as a non engineer can do with Vibe Coding, you know, Franz and I find this very reminiscent of Clay Christiane's classic disruptive innovation model. This idea that, you know, today, these Vibe Coding platforms are really best at serving these low end use cases. You know, they're not about replacing professional software developers, but they're letting non engineers build tiny little things that can still be very useful to them that weren't actually worth having professional engineers, tackle before. And so, no better representation of this than something we call the Lemkin Scale of Biped Coding. Alright. So if if you know Jason Lemkin, you know, the founder of SaaStr, a very, energetic personality. You know, this summer, he got very excited about using these Vibe Coding platforms to build apps for his own business. And, he he documented on LinkedIn, you know, as he was going through this process. And he starts out, he's incredibly enthusiastic. Oh, this is gonna be amazing. I can just build all this stuff myself. You know, a few days in, you know, you you start to see the cracks. It's like, okay. Well, yeah. Now some this thing is getting larger and more complex. I fix one thing here, it breaks something else there. And after a couple weeks or so of this, he finally published a great post on LinkedIn. Just like, okay. There are things you can do with live coding as a non engineer that are pretty safe and easy. And these were his green light things, like basic little web apps or, internal dashboards, maybe your own little custom workflow for your team, maybe a prototype that you wanna be able to show them professional developers to build something on your behalf. These are generally quite safe and a lot of, you know, low end use cases unlocked. Yellow light is, okay, you start dealing with actual customer data, you wanna be much more careful about that, compliance issues. Orange light, where it starts getting into just more complex programs that, to be honest, if you're just not an engineer, this ability to even, like, manage the complexity of that gets challenging. And then this red light zone is, like, for the love of God, don't, like, you know, try and, like, recreate Salesforce or something like this through a prop. So we laid this out on a scale from, you know, one being the green zone and 10 being the red zone, and the Lumpkin scale five coding. And we'll see how this shifts over time. But it seems like this really did a lot to, yeah, your hack hack stack model. Yeah. And and and maybe not surprisingly, but the the thing is this. We spoke about Lovable. One of the other tools is called Replic, and the founder of Replic said, you know, use the our, pipe coating tool not as a senior developer or an architect or an engineer. Use it as a junior developer, and and that is so cool because how often you as a marketer or marketing ops person thought, what if I could just make it work and show as a prototype and then rebuild it? Well, that's exactly what is happening. So it really helps with experimentation. And in the Hackback stack, we're basically showing a similar or maybe exactly the same type of dimension and scale of coding. So you you need a lot of experimentation, flexibility, agility to prototype something really rapidly if you hack. You don't want to migrate everything into your main core stack and then reconfigure everything and to find out, it doesn't work, flying doesn't bite, let's drop it. That's a pain for IT, for your stack and also for yourself, your resources. There's a huge dependency on the on the IT department who's, you know, making sure that your stack on the other end of the spectrum is really performing, scalable, no exceptions. We have to exploit current customer journeys and, revenue in the best way possible at the lowest cost. No surprises, please. FICO coding is more probabilistic, so more surprises, and that's not what you want to have in your stack. And this is where I think the deterministic SaaS type of things are at the moment really, really important. So this scale is really from, you know, on the left hand side, hacking, then we start to package it, making sure we take out all the exceptions, the data points we don't need, the integrations we don't need, even the features, the buttons we don't need before we ship it or brief it to IT and to the department who is, you know, making sure that the stack is up and running twenty four seven entire year round. And there you have it. There's a big difference in how we approach it. So one is more like a marathon, number three, the stack, and the other one is more like small sprints. And that also comes with the territory of the laboratory is working differently than the factory. And in between, we have the packaging. The people that clean up and make sure, yeah, this is the bare minimum of what we need, and then from there, we can migrate to to the core stack. Yeah. So that's where, we came with the idea of, yeah, there's a laboratory where, yeah, you could say it's more marketing, maybe sales, maybe product management, and on the other end of the spectrum, there's a factory. You have a lot of IT engineers, architects, and they are both right. And so often, we see there's a debate like, no. You shouldn't have, point solutions. Well, point solutions are maybe really needed in laboratory situations and make sure you deprecate them. Deprecation is not something we do in a factory because it has to be up and up. So these are different mindsets, different principles, design principles, if you like. And and here are some of those principles and, stack roles that we define. In a laboratory, you need to experiment, you need to be agile. You want to find new journeys because if you don't find new journeys today, in five years, your brand is irrelevant. So it's so important to have a laboratory function with a mandate and with a budget, and all too often we don't see that. And here it comes. If it comes to the balance sheet, you know, the CFO sees what the total cost of ownership of a tool is of a license fee, but that is all in the territory of the factory. Missed opportunity is something that it sits in the laboratory, and the missed opportunity, you will not find on the balance sheet. So this is why we have to make sure there are two roles, and they're both official with a mandate and a budget. Alright. Well, I know we're almost out of time here, but we've now got into, like, our favorite section of this about being the change agent. And what's at the core of that? The change agent, that is you. We have to embrace new skills. It's so interesting. Are agents really autonomous? I mean, agency refers to the word autonomy. No. What we see in our research from our, survey is that humans are always in the loop. They all decide or approve or they maybe can reverse or they review stuff that agents have been doing. So we're totally not out of the loop, but we do need new skills to moderate rather than to operate. And I think that brings a new new game altogether. Well, this is one of the things we've been talking a lot about. I mean, if there's one phrase, I think everyone here in the audience has heard this year, it's, oh, you gotta do more with less, you know. And there's nothing inherently wrong with that, you know, this idea of the ability to use AI to, yeah, improve the efficiency, of what we're currently doing, it makes a lot of sense. But it only takes you so far. You can squeeze a certain amount of profit out of that, you know. But one of the things to keep in mind is this efficiency gain through AI, it's not a differentiator. You know? Every single business on the planet is going to get the benefit of that efficiency of leveraging these AI tools. To really think about, like, how do you differentiate your business? How do you create competitive advantage? You gotta move beyond thinking about efficiency to be able to think about the effectiveness. What can we do that just wasn't even possible before? What can we create, you know, that's something beyond what our capabilities were in the pre AI era? And to be honest, even if doing those sorts of things ends up requiring more investment, that's okay, because at the end of the day, you know, it's the relationship between the investment and the revenue. You know, and if you can, like, dramatically increase the revenue, through even a modest increase, you know, in the investment and the cost, if that's like maximizing your profit, that is absolutely the way to go. And I think we're very excited to see more of that moving forward here into 2026. Yeah. And if you look at what we discussed earlier, what you discussed earlier, like, how many agents on average do we have? Then today, it's mostly agents for marketers, and that makes a lot of sense. We have to improve, our productivity, our efficiency, but the opportunities are endless almost at the agents of customers. This is where we really think the disruption will go. And, yeah, maybe the way to, serve those, agents of customers is something that companies will explore in the next, I don't know, couple of years, at least in 2026, I hope. So we have to change our skill set, and, here's an overview of how you could look at marketing operations. Right? We are, engaging in technology and data. The next step would be more on the people and the process, so it's more on use cases and take it from there. And then finally, becoming value engineers. And before we explain what that really is and why it's important, let's look at today. This is mostly where our time is spent today. And we're trying to solve a lot of problems with technology and data, but that's not always the right way to go. We're not saying don't use technology and data, but use it in the right way. And what is a smart way of doing it? Well, why not focus on the company results, the revenue? 80% is repeatable revenue. Right? And then 20% is one off revenue. If we look at how we focus all our technology, our data, our content, the pyramid is flipped. And maybe with a, minority of our all our content, technology, and and data, We can serve all those repeatable customer journeys that are profitable and bring 80% of the revenue. And this is a very important mental model. And in that 80%, we can see that they're not a gazillion customer journeys personas and try to cover everything and every single point. I sometimes make this really bad joke where I say, if you try to be at every touch point for every single customer, that's that's not marketing. It's more stalking, and that's not okay. So what you see here is in our research, we see that you need three to five really solid business cases that together bring, let's say, the 80% of the Pareto, 80% of your revenue. And maybe some of you remember the Stacking, award submission by Philips a couple of years ago. It shows one landing page and then how it in all the data, all the content, and where it comes from. But what it really shows is that they focus on their core business because razors for men was like, I don't know, I'm making the number up here, 30% of their revenue. And then if they have fixed that, maybe they go into raises for women and then take it to the next step. So this is how you really, have a very steep learning curve where you learn these are all the rules for one product and market, and then we can see and migrate to other areas and see what the differences are. And this is a very, very smart way of unpacking your revenue and supporting it with the right daytime with the right content because then you have a learning curve where you know exactly, okay. It's not yes or no. We always need, this tool, always need this, that. It's when then. So raise this from end, raise for it, etcetera. You can be very, very smart about this. The big thing is you save a lot of resources because you don't have to boil the ocean of data, technology, and content, but you know exactly. You reverse it. Go like, okay. We follow revenue. We follow the money and reverse engineer in our tooling, in our stack, in our content. Awesome. Well, there's a lot more, on all of these topics here in the report. Again, hopefully, you've downloaded, but, yeah, if you haven't yet, it should be in that doc tab, you know, in your right bar here in Goldcast. Okay. Coming up next, Franz and I are gonna take a really super quick break, and then we're gonna dive into candid interviews with our seven sponsors. We're gonna talk to Rebecca Corless, the VP of Marketing at GrowthBlue, about Minding the Gap, Why AI Success Starts with Data. We'll talk with, Tasia Spannohar, the Co CEO of Hightouch, on AI agents and the future of marketing workflows. An incredible conversation coming up here with Alexis Carcant, of Intuit Mailchimp on AI and the mid market marketing revolution. Intuit had done a pretty big study of the mid market with the firm WARC, and Alexis will share some of the results from that with you. Patrick Harrington, who is literally a rocket scientist as far as I'm concerned, the head of AI and ML at MetaRouter is gonna talk to us about AI and the first mile and digital experiences. We then have our returning champion, Sarah Fats, senior director of progress. Yeah. Dive into the the change agent, the human dimension of AI and MarTech. We then go and chat with Jonathan Moran, our long time friend, in this work, having discussion about, hey. With AI, more is not necessarily better. Better is better. What does that look like? What does that mean? And then we close out with Rafa Flores, the chief product officer of Pressure Data, for discussion about the future of AI and CVPs in marketing. So whoo, it's gonna I hope you enjoyed this hour so far. We got a couple more hours of some, like, really entertaining content and nice discussions with folks. My dear friend, Franz, thank you again as always for the collaboration on this. Really excited about what we produced. Thank you too, and I'm really looking forward to the next six months. There is so much happening. There's so much work to do, but I'm really happy we brought it to, to this report and and webinar and these insights. Thank you so much for joining. Alright. Thank you, everyone. We'll see you at these interviews next. Bye.