Why Marketing Automation Isn’t Marketing Intelligence with Duarte Garrido, Co-Founder of Dojo AI | Ep. 443

0:00Did you know 21 % of media time spent in any given day is with audio, but it only gets 5 % of media budgets. This represents a huge gap in your brand and media strategy. Learn about the largest addressable audio ecosystem at www.seriousxmmedia.com. slash Daniel.

0:27Welcome to the Marketing Millennials, the no BS marketing podcast. I'm Daniel Murray and join me for unfiltered conversations with the brains behind marketing's coolest companies. The one request I tell our guests stories or it didn't happen. Get ready to turn the f*** up. We are back with another episode of the Market Millennials Podcast. I have DG here, founder of Dojo AI, but also former marketer. So we have a marketer, but I told him I think founders have to be a little bit marketer. So I still consider him a marketer. At least for this conversation, he's a marketer. Well, welcome to the pod, DG.

1:16Thanks, Daniel. Thanks for having me. I'm very much still consider myself a marketer. Well, right now, as a founder of an early stage company, I kind of consider myself everything. But marketing is still deeply ingrained in me. Did you want to give you a background? Like, why did you decide to move from marketing to founder? And then we'll go into the conversation today. Absolutely. So a little bit about me. My background, as I said, is in marketing. I did run marketing teams for over 10 years in Fortune 100s. So Coca-Cola, Standard Chartered, Sky, then eventually got really bored of big corporate, decided to move into the world of startups and scale-ups.

1:55I was kind of infatuated with tech. I wanted to give it a go. I also had this sort of this dream that if I only moved to a nimbler, smaller organization, things would be much easier. You'd be able to experiment all the time. Marketing would have to work because you'd be held accountable for it. There'd be less politics. and I think marketing in itself as a function would be kind of solved. And there was a big slap in the face when I started working with startups and scale-ups and realized that for different reasons, but marketing still is a function in crises. I found that in big companies and small companies alike.

2:32And at the time, I kind of set myself on this mission of trying to understand why CMOs have the shortest tenure in the C-suite, why marketers are constantly frustrated, why MarTech has pivoted from 150 tools to 15 ,000 in just 10 years, and why there's so much bad data, so much lack of accountability and impact, or at least struggle to prove impact. And my realization at the time was that it has very little to do with the profession in itself, very little to do with the marketers, with the skill set, even with organizational challenges. It has all to do with technology. I think that marketers have been hurt by technology for the last decade, at least.

3:26You know, when digital marketing came about, we were promised magical attribution. We were promised, you know, ROI on every dollar spent. And we were promised really good, strong data. And what we got was a lot of data, but not necessarily better or good. And it actually, we've been held hostage by the platforms and the proliferation of platforms like Google, like Meta, like LinkedIn, like Twitter, like all these places where we spend our money, our time, our content, our resources in, they basically, they started to dictate what marketing is. marketing used to be about the product and the price and the positioning and the placement and it used to be that marketers were brought into an organization to understand the consumer and the buyer and the other side of the product or the service they were brought in because they were the closest to that audience or to that consumer and that has been slowly stripped away from us and we became essentially kind of paid media managers, SEO managers, and some digital marketers.

4:42And the platforms have dictated how we work, and they worked so poorly, and marketing became so complex. You know, different platforms call the same thing different things, they call different things the same thing. So marketing data has become basically uncrunchable. So the proliferation of middleware in MarTech has come in to try and solve that challenge, right? We had a lot of tools that came in and said, don't worry, we know that, you know, you can't manage 20 channels at the same time, but we promise you better data. We promise you better data. We promise you better data. And all we have is more middleware, more middleware.

5:22And then the MarTech tech became even more fragmented and marketers' job became even more kind of operational, almost like button click pushers, right? uncreative, unstrategic, uncritical. And at the time I was watching, this was two years ago, I was watching the dawn of this new age of AI. Argentic was just sort of peaking. And I thought, maybe this is a chance for us to take a step back. Maybe this is a chance for us to, you know, and every founder will say, oh, we're rethinking X from first principles. and they almost never do. But I really wanted to do that. I wanted to rethink marketing from first principles.

6:06I wanted to rethink MarTech from first principles and go back to what actually matters. And so I joined hands with my co-founder now who was for 20 years the head of AI and data for a couple of unicorns. So building agentic systems, complex data layers. At the time I reached out to him, I said, there's clearly a problem here to be solved. and he said, get me in a room, get me meetings with over 100 marketers. Let's interview them for a couple of months and see if I get convinced that that's the case. And so that's what we did. And sort of late 2024, we started to over 100 marketers and it was so blatant that there was a problem here that needed solving that for me, my quest became to take automation, the word automation out of marketing completely.

7:02And that's the origin story of Dojo. Today, we're a post-seed pre-series A company. We've raised 7 million in a 30 million valuation. We have over 150 corporate customers around the world, particularly the US and the UK. And we're growing at a rate of 20 % month on month. So it's going well, but I'm still wearing a lot of hats. I like what you said, and I resonated on what you said a lot, because I do think a lot of marketing teams are stifled by the tools that they have. So they can only do certain campaigns around tools. So when they come up with a campaign, they are saying, like, can my tools execute this campaign or not?

7:51Instead of thinking about this is the campaign I want to do and not think about if the tools can or cannot do that. And I think that is something that's been happening for at least five years in marketing where they're like, I want to run this sort of, I know you don't like, but like there used to be, I want to run this sort of automation. can I execute it can the tool I have executed or do I need to find a new tool and then you layer a tool that can maybe do it but then you only need that tool for one campaign so then you bought another tool and then you just like stack your tools I know you want to get rid of the word automation so when you hear someone say like I am doing automation versus or AI automation and then AI intelligence, sometimes people like use that word entertainably, like automation and intelligence is very interchangeable.

8:43So can you give me an example where like automation gives you one thing and intelligence gives you another? Like what is the difference between those both? Absolutely. So it's an interesting one because there are a few concepts that I think we are hostage to. One of it is automation, the other one is campaign, and you kind of touched on both of them. I'll get into campaigns in a second, but let's tackle your question first. So automation versus intelligence, what is the actual difference? So automation, you are setting something up not to reach a specific goal, but to just push out things independent of what the context is, independent of what the impact is, right?

9:31So automation is just grab this manual task that I should be doing by hand and just make it to run automatically in the background. Now, with AI, I thought we were going to get away from that and into real intelligence and context, which is, okay, we want to get from point A to point B. What is the best way? And that kind of circumvents automation because it's not about a specific channel. It's not about a set budget. It's not about a set of creatives. It's not about a specific audience. It is about thinking on the fly, what are the best decisions I can make? Copy adaptations, creative adaptations, budget reallocations, channel picking, whatever, to get to this specific goal, right?

10:20But what happened with AI was actually the opposite. And I'm still today a little bit shocked as to kind of how this turned out to be. Because what people started doing with AI was building more of the same crap they were already building before. Right? It's like workflows, automated workflows, now with AI, which is even worse than what we had before. because you have these deterministic workflows, but now you can do like 10x as much, 10x faster, right? You're not getting better results. You're just putting more slop into your whole marketing operation, into all of your processes, into your team.

11:07And there's so much busy work. With all of the marketing teams that I've worked in, I saw so much wasted busy work, right? So many marketers that spend their days doing things that do not move the needle. And because they are constricted by the technology that surrounds them. So that's what we're kind of taking a step back and rethinking. The concept of a campaign, right? You set a budget, you set a channel, you set an audience, you set a creative, you set a copy, and you run. The campaign has become an end in and of itself for most marketing teams, especially enterprise teams. Most enterprise teams that I've worked with, they run campaigns and they see it as facta complete no matter where, it doesn't matter if the campaign actually delivers the result they want or not.

12:03The fact that they run the campaign is already work done. We are rethinking the concept of a campaign because now with Agentec, what you can do is you can take the context, take the intelligence, take the real life data, both structured and unstructured, your channel data, but also the data that surrounds your brand, your environmental data, as a person who works with me likes to put it. I quite like that term, which is, you know, what is being said about your brand? What are your competitors doing? What are the trends surrounding your market? And all of that should be taken into consideration when you're making campaign decisions.

12:42So right now with Agentic and a strong data foundation, you can create loops, right? You can make, and this concept already exists in engineering, which is a RALF loop. So basically agents that run in a loop until they get to a specific goal. And they will decide on the fly what it is that they need to do until they reach that goal. Now, what we don't want is to turn that into automation in a sense that we don't want to delegate decisions to AI. AI is very good at some things, and we are now learning as an industry, but actually we're learning as a society that AI really sucks at many other things that we thought it was good at.

13:24So if you want to do great data analysis, you need a non-AI layer on top of it because you need some deterministic system to actually analyze data. If you want to do project management, if you want to build a product, you need a classic SaaS layer, okay? The prompting, the endless prompting and endless wall of text isn't going to be a SaaS product in and of itself. So we are basically putting all of this together, the AI, the context, the live data, the process, the project management, to bring the human to be truly in the loop in these agentic loops. right? The human will coordinate a system that is not autonomous, but that it is intelligent.

14:15So it will run, it will suggest things, it will suggest changes, it will suggest opportunities, it will give you ideas. But you as a human, as a marketing team, you maintain the judgment, you maintain the decisioning, you maintain the critical thinking. But most importantly, you maintain the creativity. Because another thing that AI really sucks at is creativity. Everyone rushed when sort of Gen AI took flight. Everyone rushed into creative workflows. Oh, AI was going to replace copywriters. AI was going to replace designers. AI was going to replace... And actually, we have a real problem now at hands with AI Slop.

15:00A real problem. and if anything, it's detrimental to trust, it's making us lazy and it's taking away from marketers what marketers are actually very good at but most importantly, what they really enjoy doing. We cannot let that happen, right? So the prompt enter and creative pops out on the other end needs to stop, right? Creativity needs to be in the hands of marketers. And we believe that deeply. I also want to know, like, what is the checklist of, like, business data or data in the company that you need before you can trust the system with intelligence to make a correct marketing decision? Because I think that's, like, one of the things, like, what are the inputs that I need to make sure that I can trust that this is making the correct decision for my business?

15:58Because just like you said, like the end output is slop. Usually it's because there's inputs that are slop. Like whoever's writing, like if a good writer uses AI, I believe that like the end product will become something good because they're also going to add their touch and anti-laziness to it. But if a bad writer or lazy writer goes into it, same as like any decision, if someone doesn't understand data, and they use it to like come up with a data decision. They don't understand what this data is or where it's coming from. I still think it's going to make a bad decision, but I want to know what data you think or like the checklist that marketers need before they could trust AI to make an intelligent decision.

16:49How much of your day do you spend listening to content? For me, it's constant. Music on a wall. Whatever podcast my dear wife is blasting in the morning, in the car, YouTube music for my toddler's latest favorite song. Audio is with you when no other medium is. That goes for your audience too. Sirius XM Media is the single largest audio ecosystem in the US, reaching 9 in 10 Americans. Time to work audio in your advertising plan. Learn more about the largest addressable audio ecosystem at www.seriousxmmedia.com. So I think that's a really good question. I don't think that the weight should be put on the marketer or the user of AI.

17:42And the reason is, I actually don't think there will be much of a difference of a good copywriter versus a bad copywriter using the same LLM for copy. I think the difference will be marginal. Sure, there's endless prompting you can make to try and make it better, but there's a ceiling, and the ceiling is actually quite low in copy. I can smell clod copy 10 miles away. The same thing for ChatGPT now, the same thing for Gemini. You know the quirks. Now with the watermarking, it's even worse. like the watermarking is not some kind of hidden object that'll be in the text. It'll be the actual way these LLMs write.

18:24So they will write in a way that is so formulaic that it becomes watermarked. So that means that it doesn't matter how good you are. If you are using an LLM to create copy, your copy is going to be crap. I was more saying that like, Like, they have to have, a writer needs to put in their input and edit out. They need both. They need to be on both sides of that. Like, they can't just, like, put in what they think would be good writing and then edit output. They need to have the touch at the end to know what taste looks like. I think, like, that's a problem is that a bad writer wouldn't be able to tell, like, it's bad writing.

19:08Yes. Or like someone who knows AI wouldn't know that. Someone like you who could smell AI knows it's AI, but like someone who can't smell AI would just think it's good writing. So I just think that there's those things where you need that like taste and knowledge of like the system to be able to understand it. But I do agree that if prompting your way to good writing will never create good writing. No, and I think you're right. And I actually didn't answer your question, which is what is the data that is needed, the input that is needed for a company to make AI work for it? And in a word, that's context, right?

19:51And that's what we've built with Dojo is on top of the data layer, we have a context layer. And this context layer is essentially a curated semantic model of your brand and your marketing operation. and it has all these data points that we've trained Dojo to go and get for you. So they can come from your own channels, from your paid channels, from your organic socials, from your website, from your CRM, from all of that, they can come from our own data sources around SEO, AEO, traffic, competitor data and they can come from public available data, Reddit, you know, so the voice of customers, the market movements, the competitor movements, the brand awareness and equity piece, the proxies for NPS.

20:42All of that is data that is useful for the agents to build this curated semantic model of your brand and to make Dojo AI your Dojo, right? And work for your brand and know exactly what it is. But then there's a layer of it that is yours to input. yours as a user, yours as a company. There is always a layer of customization. And we've built Dojo specifically for that. So that you can come in, the semantic model is built, the context layer is built, right? But everything that you teach Dojo, Dojo is built on what we call a closed loop system. So it means it never forgets anything and every interaction it has with you, every action, every campaign that it runs, every feedback point that you give it, it will learn from it won't forget and apply it to your whole account to all of your users so suddenly for an organization or for a marketing team you have this second brain that keeps getting bigger keeps getting better right and that can actually help you identify opportunities where you couldn't before because it has intelligence instead of automation and it can spot threats a mile away again because it knows what to look for And this is the real pivot in marketing that we're about to see.

22:00It's less about producing more of whatever. It's less about integration. It really is about intelligence. And the word intelligence here has been thrown around too much. But intelligence has actually been lacking in marketing for a really long time. I want to use the classic paid media example. So if Meta is basically already automating, targeting, bidding, creative testing, optimization. What would that intelligence layer do that the algorithms can't do? That's excellent because that's part of the problem that I'm seeing right now, is that the platforms are forcing you to put them on autopilot, especially on the Google side.

22:50They're now pushing everything into AI Max. That means that PPC managers are kind of almost, they've got their hands tied in many ways, especially when it comes to broad search. And the reality is that Google only has Google context. The only context that it has is your historical campaign context within Google. Meta only has your meta context, right? So all these decisions that these platforms are making on autopilot, they are missing context. They are missing intelligence. They are missing data. They are missing information. And now they're removing the user. They're removing the user who knows more than they do.

23:33Obviously, the user knows more than they do because it's actually privy to much more data, much more channels, much more context than these platforms are. so that's where an intelligent marketing system like dojo can come in and go like no no you don't need to put this on autopilot you can manage it through a centralized system where you're actually in control still but yes you have all the context all of the insights all of the data yeah because you still run into with all these tools or like platforms the walled garden issue where like like Google is definitely not sharing data with Meta. Meta is definitely not sharing data with if you're doing Snapchat or Amazon testing or whatever your brand's doing, it's not sharing that data.

24:23So the context is only like, okay, you ran this campaign in Meta before. Let's just optimize it a little better because it's, but not like, oh, I'm seeing this message of testing working on that platform better or this audience is working here and you should move this audience here as well. It doesn't know that context because, or like this is the ICP you should be targeting versus the ICP. So I think that's the wall garden approach. But I also wanted to ask you, a lot of companies are trying to like automate tasks right now, but like what's a task you see marketers trying to automate like outside the platform thing, but they should really like rethink that approach and just not automate it and do something completely different.

25:16I actually go back to cadences in general. I hate them. I've always hated cadences. I spent a lot of my marketing days building them and they never worked as they should. And I always looked at them distrustfully. I was always like, I remember my initial sort of five years in marketing where I started, I wanted to be a kind of a polymath and be kind of more horizontal in marketing. So I wanted to learn all of the platforms and all of the channels. And I started building these cadences, like if this happens and that happens, if that happens and this happens. And all of it seemed to me deterministic and just not the way the world works, not the way marketing works, not the way people act or think or behave, right?

26:05So we've tried to simplify this because that was the only thing that automation allowed us to do, was to have triggers and then content. Trigger, content, trigger, content. And it never really worked. And then we applied that to everything all the way up to SDRing, which I kind of see as part of marketing. Sometimes it is, sometimes it isn't, but mostly go-to-market as a function is becoming a bit more compact. So the SDRs, they're building these cadences, right? And all of them are triggered by intense signals that come from bad data. And right now we're seeing a complete saturation of this. Like we get cold messages all day, every day, to a point where we don't even react to it.

26:59anymore, right? We kind of get them, like, if I get a cold message on LinkedIn, I don't read it. It might be interesting, but I immediately close it. If I see a number calling on my phone that I don't know, I immediately send it to voicemail because I know what's happening, right? So we're exhausting channels. And it's happening, the same thing is happening for paid. The same thing is happening for SEO. We're kind of driving all of these mediums to a point of saturation because we are not approaching this the right way we are approaching it in a more faster now with ai way of thinking and that's wrong-headed so i think we will there will be a moment of reckoning very soon it's a classic what happened like when people used to win because they used to know how to target better than everybody else.

27:58And then like targeting became like table stakes. And now you can't, you like, like they didn't focus on creative. They didn't focus on what actual copy. They just were like, I know how to get an ad to the right audience on this platform and let them see it. And I know how to manipulate the platform to do that. And then now it's the same thing with like, yeah, it's going to work. to a point and then everything's going to catch up and then you're going to fall behind because you didn't do the things that led you to get to that point. But I wanted to also ask you, because let's say like the classic two marketers have the same AI tools, like the smarter marketer does what to win?

28:46Like what is the better marketer does what to win? I love that. Everything AI touches eventually becomes a commodity. I believe that deeply. LLMs are already commoditized. Intelligence will become free, not cheap, free, the same way content did a while ago and data did. And now insights are also free. Like everything is becoming commoditized. So what is then the differentiator? What makes a marketer a good marketer versus a bad marketer? And that is genuinely the ability to choose what to do and the ability to think different. Marketers were always the people who thought differently, right? They were the ones who could come out with out-of-the-box ideas.

29:40AI sucks at it. It is going to regurgitate, even if it's been a very carefully curated model, It is not a tool. It is not a technology that is built for creative ideation. Because creativity actually comes from human experience, not from data. And I would actually add one thing to that. I think like AI is built on logic. Like it's supposed to give you like the logical answer. and usually the campaign that is different is something that defies what logic is. Yeah, exactly. And that's the problem. Like you can't win on a logical answer. You might get a slight left, like we've talked about a little bit of like, you might get that incremental, like a little bit better, but logic at the end of the day is not what separates like this brand versus this brand.

30:40It's someone who was like, I want to make a water brand about death. Exactly. They chose that. Logic will not tell you to make a water brand that talks about death. It just won't. That's exactly right. And isn't that exciting though? Isn't it exciting thinking about marketing in the way the madman of the 50s and 60s did? which is let's sit down in a room, have a deep understanding of the customer, have all of the data and all of the insights which today we have, but then get our best minds in the business around the table and brainstorm creatively how to puncture the noise, right? How to stay above the fray.

31:29There's going to be so much of everything, right? All the channels be so saturated that we will have to really up our game. Companies, I'll speak in the perspective of a company, of a brand, right? Whatever used to work for a brand to be heard two years ago doesn't today. And whatever works today will not work next year because everyone will be upping the volume, upping the targeting, the personalization, the everything. So marketers, human brains, will be the ones who will have to actually think of new creative solutions to put our products in front of our target customers. and that's going to be such a fun challenge it's going to be amazing and we're going to have the whole operating system that allows us to run and make good decisions on our judgment on our creativity as marketers and we will be needed more than ever by the way and it's also i do agree that like the difference between madmen and now is the the speed of which we can get results back and answers back and the level of budget to get those answers are way different than madman like you would have to put something big out there to get answers or direct mail which took time or like there's a lot of things or TV that you didn't have data back but now now you could run data and find okay this headline or these customers are saying this this and that like how could I make a campaign around like the words of my customer that could be interesting which is way different than like having to do all these focus groups and then have someone like mainly write by hand and then someone going through all those like answers and then deciding what what is the commonalities what is not the commonalities what are the words like those things are ai is good at because it takes like unstructured data and gives you like an answer quickly where then you as a marketer we We could take what we know people care about, words people use, reality TV shows they watch, whatever, and take two data points, which are new ideas really like two data points that haven't been connected before.

34:00Really, that's what it is. Yeah. But if you need data points to like take those two data points and connect them, that's why I like AI because you could get data points that say your audience like really loves this TV show and they love, they are saying this, like how could I marry those two and make a new idea? Exactly. Between those two. so and how do i and dan how do i grab those insights that i turn them into a campaign on three four channels and build a landing page and push them into it and like all of that will be possible in the blink of an eye i have a question because i think marketers are probably struggling with this because sometimes i struggle with this with ai is like how do i know like it has enough context that i've given it to make a good answer like what is like what is enough context or like that word like because people always say give it context and then it'll make a good decision like what is context i obviously i'm a little bit biased here but there is no enough context you can manually give an ai because the context window will be reduced and if you give it too much it'll forget most of it uh and if you give it too little it's too little so i mean that's mostly why we've built Dojo, right?