A DTC Strategist Thought Claude Would Replace Him. But It Just Gave Him Superpowers

The D2C podcast is brought to you by Pylethos, the performance agency behind some of the fastest growing D2C brands in the world. Creative, media, and customer journey all under one roof, performance and brand without the trade-off. Every Friday, we hand the mic to a Pylethos operator to break down what's actually working in their space right now. Want a team that treats your growth like their own? That's Pylethos. Head to Pylethos.co and now on with the show.

One of the first things that you're going to run into as a strategist is trying to figure out

what the hell is going on when you're in a world of enormous amounts of context and data. AI helps you to figure out what the hell is going on with having to read every single spreadsheet. In any organization, you can work cross-functionally without the department. But if your models don't have connection points, they don't work. I remember very clearly it was like January last year, downloaded cloud, started messing around with skills,

and I was like, "I'm dead." It is over for me. When I realized, is that it doesn't replace judgment. The hardest part about AI is actually integrating it into a world where there's legacy structures and legacy systems. This episode is brought to you by Triple Whale, the AI operating system for e-commerce. Just a quick gut check for brand owners and media buyers. Have you started your Black Friday

Cyber Monday planning yet? If the answer is not yet, you're not alone. But the brands that win

BFCM aren't the one scrambling in November. They're the ones planning right now. And here's why it matters. BFCM keeps getting more expensive. Last year, ad spend climbed faster than demanded CPMs went up, and the season stretched from a single weekend into a full month of promotions. Show up unprepared and you spend more to reach the same shoppers right when your margin matters the most. Triple Whale built a free resource to help the

BFCM hub. It's like a command center for planning your most profitable Black Friday with everything in one place. Real benchmarks from tens of thousands of brands so you know what performance should actually look like for brands just like yours. Free tools, to set goals, and forecast outcomes. Strategy guides and checklist covering ad performance retention and conversion. An AI powered tactics you can put to work before BFCM weak hits. The best part, it's completely

free, and there's no log in required. Go to triplewhale.com/bfcm26 and start building towards your most profitable Black Friday Cyber Monday today. It's all killer, no filler, and dunk

“and just finish the conversation. What were you get at what were you saying, dunking?”

I just finished a conversation. I've been, yeah, I can like crazy lately with a new strategist. And he was asking questions on how to get into the rhythm of like what doing the job is. And I told him that then like one of the number one skills of many, but one of the first things that you're going to run into as a strategist is trying to figure out what the hell is going on. And when you're in a world of enormous amounts of contacts and data, AI helps you establish

very quickly, especially when you build your own tools, to figure out what the hell is going on, with having to read every single spreadsheet manually typing things in. So enormous, it's very good as a context gather use correctly. But I think like Raph is in a really good job of building a system that can take enormous amounts of contacts, get a very clear stance using his own experience and processing all this information into a very clear distillation of, I don't know if I can say

hell on the podcast, but what the hell is going on, which is usually enormously helpful for people because it's very overwhelming a lot of the time. So let's back up. That was, we rarely jump in pre-conversation like that, but today we are chatting with, it's just a couple weeks away from

“Black Friday's ceremony, so I think we're all feeling-- I know, trying to get a podcast today from”

the pilothouse staff has been or this week has been a big challenge, so I really appreciate you guys

stepping up. And strategy is a topic that, you know, I think our audience is always interested in,

and I think it's, you guys are kind of leading strategy from different sides within the pilothouse organization and you are just sort of talking through a little bit about how you guys are operating, how you're using AI, and so yeah, so let's, let's hash it out. How would you frame it there, Raph? Yeah, I mean, I think, I think Duncan framed it really well. I feel like there's this constant struggle of, you know, all AI, I know AI, but I think maybe I'm just a general

centrist, but I believe it's somewhere in the middle. I think through Duncan's point, I think what AI does really well is it's able to compress and compute a lot of information, but I think without having a human holding the context that information becomes noise if that makes

“sense. I think internally Dave, CEO of pilothouse, mixes joke, joke where like we have to create”

cloud food, but also create human food if that makes sense. So there's like millions and thousands

but then the human job is then how do you translate that immense amount of output that cloud creates? Because it's easy to create a hundred-page back now with cloud to a 10-page overview,

five-page overview, one-page overview, to allow someone ultimately to an outcome.

Yeah, somebody did exactly the absorb the outcomes of the service. So they can drive an outcome, right, as cloud is just going to help you create output at a faster pace. But if the human can grasp it, then the output is not going to get created. And as Duncan says, you know, when we talk about this quite often, how do we slow down enough to understand, at least from a sales standpoint, the delivery workflow and vice versa. So which state Duncan the docs talk, is that

how you what you call the docs talk, yeah? It is very interesting because I am a context person and I will read through most things, especially when it comes to a strategy perspective,

like RAF is very good at creating lots of context for me to who we're up. I'll also pull that

into my knowledge base and I'll use that to tap into at any given point. Then do you use a word that RAF and I like to use often as I will craft from there. And so you're sort of jamming with

“all that context, but I don't have to remember every single detail. And I can craft down and get”

down into like one sentence, what it is that I need to deliver. So when we're talking about different sides of our work and how that measures together, RAF takes the largest data set essentially the rawest one. Bring that in. He pitches that to the client. Great pitches. Then it gets handed over to my side where there's more context is brought in. I will craft that and refine that further because I have more time and then that sets a series of decisions for our client. What do we want to

do? What's the big constraint in our way? How do we unlock that constraint? Okay, cool. We're agreed. And then we build this, then we start to blueprint out. You can see behind me, blueprint out the the systems that our teams operate within that is like a series of pipe work. But it's interesting because RAF was, I think, probably the first to unlock this challenge of strategy in scale that

“as an organization we've been sort of wrestling with because there's so much constantly you need to”

deal with, building these systems that can distill that down into something manageable, but it does need a mirror side on the other side of partnerships to receive it. I think sometimes for it to work.

And I think that's the critical part, at least, you know, I know that's all kind of nascent in

everybody talks about using AI, but at least for me, that's one of the big practical things that I've learned through this process is that an organization can work cross-functionally without the department, right? But if your models, at least one have connection points, they don't work. I'll give you an example, right? Like, let's say I have to in the sales side, identify what the client's goal is, right? And then Duncan also asked me to identify the client's

goal for the quarter, you know, because we're helping them live. If my definition of goal is different than his definition of goal, when AI is sourcing and pulling content from my section of my presentation report that talks about goal, and that doesn't match Duncan's expectation, then it's going to create confusion. And that's a simple example, right? So a big part of this process is so important to make sure that you create the object relationships on the

back end that work before both departments, or else then your knowledge based doesn't scale, because all the work that I do can't really translate into how Duncan and Anna's team works, because the objects and how we define the objects are not the same, right? So yeah, so like we don't object permanence to something. Yeah, so then as you know, Duncan and I have work through it, I feel like Duncan's analogy of, you know, the dogs need the dog, you know, like the train has

to have a station. Yeah, and that there's a subway station for like, you know, a steam railway

“track is just not, you know, going to kind of connect, right? So I think for our listeners it might”

help just to give a little context to about what actually your parts in the company are so if you stayed with us to this point, no, I, and I'm curious, and I'm just curious, I'll wrap on your side. Like you mentioned, we talked in the pre-call about how much the degree to which we're getting context with our partners now. And I think it belies this idea of a shift that Pilateau says undergone from a multi-channel marketing partner, advertising agency

to like a really fully integrated growth partner. And I think what you guys are both talking

you can't just be like run more ads, get more creative, like it's got to go down to contribution

“margin. And when you're dealing with a problem that big AI can be a massive help, both for how you”

onboard clients like in the sales process, and then also, you know, set them up for success with the delivery team. Did I summarize that well? One hundred percent, and, and to that point, there's not not a sales pitch, but, you know, on those sales goals. So, yes, I am the head of strategic growth, I support that the partnerships team, and ensuring that, you know, we said, we understand the client's problem, and we solved for it, right? So, my simple pitch and how

I introduce myself in all those calls is that our business model is simple, right? We invest

in these audits because we want to make sure that we can help you. Because ultimately, our growth

model is very simple. You grow, we grow alongside of you. We want to make sure that we're investing time in the right opportunities, right? So, we do a pretty in that process to talk in point, you know, 10, 10 business days, and we basically come out by identifying your core problems, we'll do like full unit economics analysis, and even some positioning conversation as well as a immediate media plan alongside it full technical audit. Our team

small, basically, myself, another sales executive, and then some support from the defense, but we're cranking, you know, at the height, anywhere from, I would say, 5 to 12 pitches in every two weeks,

“like all, you know, it's quite a bit. You're doing a full strategic plan on the fly, right?”

And the amazing thing, and I've shared this with you of early on this call, it's amazing how much

access we get, you know, we get cogs, we get access to the shop, if we can see all the revenue at the end, cohorts, positioning documents, access to other platforms, and I think most agencies, you know, do a very surface level scan, but exciting thing for us, I believe, with the advent of AI, we can compress the amount of time in value, we can create with the information we have. The liver value set up the context of how to start engagement on the right foot, so when

Duncan and team get the content, they have a starting point, so it's not another two to three months of onboarding and delivery. So that's really what we're trying to implement and scale, right? So through the sales process, if you're going to miss time,

we want to give you value by creating a head start for our team, so.

Duncan, you spoke to a little bit of the degree of granularity delivered at both stages here. What is the outcome when they're finished with you, Raph? Once you once you take it in all this context, what is the degree of the, of the plot, like the play, you don't get down into angles and like, you know, eggs, add angles or things like that, but what's the degree of granularity that you provide at your stage of the, like, it's a diagnosis of the main problems of the bottlenecks,

what else is the delivery there? I'll give you, like, one recent example, right? Because most of the classes we deal with are mid-market, they're going through a former transition, right? So that does your brand that recently hit nine figure, and then the growth has completely flat-line. They're like, we don't know why, you know, we haven't changed anything. So what we delivered there was that, hey, you know, you have too many customers in these three markets, so

they account 70% of your media is just going through the three markets. Nobody else is seeing it. So we were able to show them why that was the growth of our flat-lining, right? You grew, right? But because these three small markets now met us just pumping all media into that market, you have no more customers to get, right? So they're sitting, trying every tactic, adding

“more creative volume, right? So that's what we provide is that initial, what is stopping you or”

causing the issue you're seeing? That's our starting point goal. We don't have the exact solution, because at the time at this point, we have to figure, hey, they have new products launching, what we market that you're entering into, how much media, we don't have those conversations, but we can narrow down on that problem with all the data access we get to yet. I would also just just add on to that, like, pilot house as a performance marketing agency,

we are incentivized to grow your business. We just have to specialize in marketing. And so that, if you follow that logic all the way through, you will get to strategy. It's kind of boundless. Sometimes it's hard for us to draw a boundary for us, because it's like, hey, we need your revenue to go up with, like, your spend or what have you, we need to make more money. That's like, we're incentivized to do that. So then, if we weren't just continue to sell

services, then it would be incentivized just to sell based on, like, I like paid by adds, Ben, I can work with that. That's a percentage of adds, Ben. And so we came to this conclusion, it's like, hey, we cannot be bound by, like, you know, adjusting your meta-span or your Google

strategies so that we can unlock your constraint. That's what Raf just talked about. What is your

constraint? Your business constraint. And so in order to do that at scale is really quite challenging. You need a collection of quite a bit of a specialized experience, but also broad enough across

“all of the channels that we operate. And to do that is really challenging. So that's why we have”

the tip of the spear. Someone like Raf, who is, you know, probably 90% efficient in most of the core skills of e-commerce to be able to take an initial look at your business, identify that constraint, and we further refine it. To do that at scale, essentially requires AI for Raf to be able to do, like, what do you say, like, 5 to 12 every week or something like that? And every two

weeks cycles, especially in summer time, like, yeah, you're crunching, like creative strategy,

production, budgets, forecasting, new and economics, media plan, all of that, yeah, and very condensed jump of time, yeah. It's smashed in. And so we have quite a few experience people at Pilatehouse who have a lot of specialized skills, but also broad enough skills to be able to be dangerous in a lot of different areas. And the level of involvement in knowledge and context and understanding needs to be

“to such a level of degree that, like, it kind of breaks at a certain level of scale. So that's why”

AI is really, really important to empower these people. So then Raf's building systems on, he's essentially doing consulting, like, an initial consulting tap on the sales side, and then that gets merged into us to carry through. And to Duncan's comment on operationalizing and figuring

out what needs to happen while you're running the account, I think it's critical because when we

do that analysis, there's many issues, right? Like, but through the pitch process, we're just pointing at the one we think is the most important one, right? So to Duncan's point, it's such a rage, like it could be creative, it could be positioning. It could be a volume thing, it could be a immediate distribution. In the pitch, we're just honing in on one to be like, hey, this is where this is what's stopping from hitting your goals. But at Duncan's point, the analysis done on everything

else. So then they can start shifting, you know, they solve the big thing. Now, we're going to move to the next thing that we need to solve. Maybe it's a creative volume issue, right? But because we have that knowledge base, and we've been trying to work together to make sure that our systems have the same object permanence or like, we're in object definitions, then Duncan can point and figure out what the next issues are going to be, or update them, you know, with new information,

as they flow through. Yeah. But it gives kind of the teams of running start. What are we building this on? Is this a cloud project? Is this a HATGPT? What where is this? For prior Terry software platform. Oh, for prior to that. To every two for me, just folks are in front of both of us. I think use primary cloud, and we share our skill files back and forth as well, just to keep those things updated. Make sure context is the same.

Yeah, Jeff just switched to chatGPT because he likes the way the agents works better. And I think

“that's what Andrew Wilkinson uses. And I was like, oh, because I've like really gone all in on”

cloud, and just to know that he's like, he thinks it's better over there on. It's front of your LLM. It's like, everyone, it's becoming the way people do like every day. I'm connecting it to a new platform. I didn't realize I could connect it to and realizing that I don't have to leave the terminal. Yeah. That much, right? Yeah. That's pretty, which is interesting. And to your point around proprietary, I think that thing that's more important, at least, you know, operating and building

this alongside with Duncan, from my perspective, the proprietary thing is the object definition and relationship building between files, folders, agents, and skills. I think that's the thing that's most difficult, right? Because it's even, it's simple things, right? Like, let's say you have an agenda document or report that you want automated. Like, if the objects are not consistent between departments and the naming structure of the folder or the document is not consistent,

then the agent can't reference or pull the right document, right? So the thing that is the most complex is designing the object relationships and object definitions and driving consistency across the partners. That's what the proprietary component. Did the agent diagnose the in this hypothetical case where the people were oversaturating their their current markets? Was that something that a human diagnosed or was that something that the AI contextual machine died? No, the human

diagnosis in and then I have enough experience to pull a set of anywhere from 8 to 14 or 15 reports

I have a hunch, you know, I pull a frequency report. I can see that we have 140 frequency in a specific area. I've seen this before, most of the time, that means that you're over investing in one place. Then I query my knowledge base. Hey, is this happening? Can you show me, validate it? And then I have

a working session basically with my knowledge base. Another example, client came to us and they're like,

oh, my AOV started to drop, you know, from about 40 and it's now to 32. And then I have some questions and then she told me, okay, you know, our main product used to be 40 dollars, which makes sense why their review was 40, but introduced 20 dollar product. And then I was like, okay, do you guys said budgets by merge volume, like product volume? And there, no, well, what's meta going to do? It's going to naturally spend against a thing that's cheaper because it has

higher volume frequency. So I've seen that. So then I query, I'm like, hey, how has spend split between this product and this product between this time and this time? Paul's a report. I might

“be wrong. Then my hypothesis is not right, right? So it's understanding the box, you need to build,”

and then as you get little bits of info, understanding what you can query. Hunches. Hunches are decidedly human. Yeah, exactly. Yeah. This is the interesting thing about building this is that I remember very clearly, it was like January last year, downloaded cloud, started messing around with skills, and I was like, I'm dead. This, it is over for me. I've been building for quite a long time in-cloud. And what I realized is that what this does and doesn't do, it doesn't replace judgment.

It allows you to have higher leverage on the judgment than you've accumulated over time. And like, that is hard one, like blood sweat and tears. And it's very interesting because it allows you to execute your judgment at faster cycles at scale, instead of fiddling around with building spreadsheets and pulling reports and things like that. You can just actually, at the human, it really is true that the human is elevated to be higher leverage. But the judgment is just

“as important there because there's still stuff you do not have data on stuff that hasn't”

happened yet. And so like it's always a backward mirror. You can look at that. But the judgment

is actually really, really important. Exactly. Which is what people come to pilot house for. We have all these players who have just been absolutely in the trenches battling it out, that accumulate the judgment, that then they can deploy at higher scale. It's still something like, it's like 50% or 50% or 60% of pilot house employees are like on Vancouver Island or in British Columbia, like in in Victoria and there are people. There's so many people that have been there for

years. Like I love the dichotomy we're talking about today of because agencies have always and I hopefully always will be deeply human businesses based on human capital. And so it's interesting talking about all these ways that I'm using AI and people are building these AI systems. But I think it's really important that it isn't in this pessimistic way where we're replacing ourselves, it's where you're elevating. So you have that opportunity to really elevate, right?

Yeah. And another example of that is, you know, a nine-figure brand that we work with. We rolled out this regionalized market maturity model. It's a strategic approach that I've used in the past. General takes me about six months to do with a data scientist, right? Dave and I ran I think four workshops across two weeks and we did all of the technical data processing ourselves through cloud. The data sheet was 178 megabytes. So it's a ton of data. And because it's a delivery,

you know, normally to do that. You have to like work through somebody who's highly technical

“to get by the you have to coach them on what good looks like because you can't do it yourself, right?”

But through this because you have to judge them in the context what data you want to think the only thing missing is how do you mechanically stitch together? Yeah. And like yeah, we work directly with the CEO. It passed through their finance CFO as well as all the numbers were correct because we have to business context, right? So it allowed us to do a six-month job in

two weeks. Yeah. Amazing. I had a really interesting conversation this weekend with an agency owner

friend of mine who actually owns multiple different businesses that are all kind of in the area of he's approaching, you know, approaching nine figures in size across these organizations. And he was sort of saying he's so invested in like what SaaS and what the way anyone initiates the digital world becomes in the next five years. He basically was saying he's like my goal is with the companies that I own is that out of these companies, there will come one new kind of the new kind of company

that will be like the agent to agent middleware piece between people's chat terminals and the

there's the pre-Amazon days and there's the post-Amazon days and as soon as Amazon came around

“Amazon changed the whole context for what it means to shop online. And everything since then”

is sort of compared and held in the frame of what Amazon did. His theory is that LLMs and agents and personal agents within the next five years get to that point where it's like so much gets computed on the other side of people's agents that all agent all services, all software tools need to almost be to be configured to exist in that like the other side of the frontier LLMs. Yeah, I also agree, but to Duncan's point, I will lay on the human element. I will I will say

I had a very similar journey when I started using cloud like Duncan did or like I was outputting over output but then what ends up happening was I got lost in the output.

Right, and then what I realized is the limiting factor is the ability for the person on the

other side whether it's the creator or the receiver of what is created to absorb the info. And if a human can't absorb the info, action's not going to happen. Do you know what it means? So like it's you can create at pace but then if the receiver, it gave the human on the other side can filter and create decisions or actions. Then it's not going to be useful. All those outputs are not going to be used, you know. It kind of reminded me of you know what said all that if you want

“to go fast, you go alone, if you want to go far, you go together. That is the human dynamic, right?”

And for me, output is information but if the actor, which is human at the end of the day, can absorb it, then there's no point in the output. Unless their agent can absorb it for them, you know what's happening. So who's going to chew it? That is exactly it. The speed and scale and it's like, if I can't, because like we, let's just think about it as like, I'll also just say, like if there's a human in the loop, the human needs to be able to absorb it.

That's just it. And like in your course, like, well, if you just don't have a human in the loop, then that doesn't become a problem. Or just on the other side of that loop, if you don't be in, like, it could be that agent like exactly how that human absorbs it. That agents in a best position to know how to help that person absorb it, their own agent. And like if you had a, if you had a, you know, an AI pizza making robot like that pizza making robot limit would not be how

many pizzas that can make in a day of you how many pizzas can, it's human owner, he's kind of,

yes, he's fancy. Yeah, he's fancy. And I think that's always going to be the limiting factor

of that automation. Because, you know, the other thing I've learned by design is like you almost have to be a bit of an engineer, you know, have version control, fuller file, hygiene systems.

“You have to build all that right because ultimately software, right? So we've clouded updates.”

If everything is fully piped and automated and the human has no idea what's happening, when software update happens, your system's going to break so that if you don't, if the human doesn't know what's actually happening there, how do you maintain the system? Do you know how water is purified? No. There's all, we, our whole society is based on not understanding or it's highly specialized knowledge too. Yeah, exactly. Sure. And I think like, if it, like, there's the realm here where like,

there's, there's legacy structures and legacy jobs. So the hardest part about AI is actually integrating it into a world where there's legacy structures and legacy systems, that's where it becomes challenging. If you wipe the whole thing clean, it might be easier to just build fresh. But like, I would say that I think everybody has this sort of experience. I had a very similar experience where I was building building building. I get lost in the thing. Then you have to have a re-learned

head of structure in a way that you can understand. And I still get caught up where I'm like, I could get clawed to do this and I frequently get get my AI agents to build me something. And I'm like, that can't be right. And I challenge challenge challenge and I spend probably more time trying to figure out how it did it. And then I realized it was right to begin with, technically, even though it seemed wrong. And it burns up a whole bunch of time, too. So like, it's very

interesting in a way because have you seen the movie Fantasia? Not for a long time, but I've heard this analogy of the brooms. The brooms, you know, Mickey Mouse, he steals, he steals the wizard hat and then he starts doing magic. And then he brings the broom to life so that the broom can carry water for him. And he makes a hundred brooms and then he drips, he almost drowns because they, the water overflows and the other, yeah. And then he has to kill all the brooms. And then the

master comes back, he's like, you shouldn't have been messing with the magic hat. And I feel like that is sort of the experience everybody has when they start building with this thing because it seems, it's a genie with seemingly endless unless you run out of tokens with is it can grant and then you're like, okay, I do got to be careful with what wishes I, I make because they actually could go sideways and take more time. So monkey-poss situation. Monkey-poss, it's really exciting.

I think you're creative person and then build once well, yeah, I would say. Well, yeah, sure,

“I mean, in the marketing world, whatever you want to call that, but I like to build. I like to make”

new things. It's a very exciting, like, world to live in. I'm much preferent. And so it's, it's, it's, it's really cool. It's really cool. And it means we can just do more of the fun stuff, which is like that strategy tier to be at, like, at the cycles of decision making and probably identification were with and through that. And it's, it's really, really cool. And I don't have to be

an expert in Excel. Yeah. Yeah, it's, it's amazing. Yeah. I'm, I'm just going to share, like,

one, one last example, too, like a, even, like, on research studies, right? Like, um, if let's say you do a big focus group that you pay a million dollars for, so you're testing a product and hearing out people are reacting, right? That's like so inaccessible to people, but like now, even in our audit process, we can do that. It's just called a semantic mapping, right? So you get people to react to a product, record what they say, and then see what words get grouped over time. But

somebody's like manually coding, you know, and mapping that, you know, right? And 2,000 can point is so fun to be creative now, because if you understand how that type of research works, you can leverage AI to do the semantic mapping. So now I can do a multi-million dollar study

by myself, you know, I don't need to hire people, you know, to code, summarize,

“start grouping, you know, and I mean, so I just unlocks the amount of creativity that you have to make.”

Yeah, I was following this guy. I was talking teaching people how to make stock tips based on like learning to train their AI to look at satellite photography of of these like tech companies parking lots, and like, you know, all these different factors about like, well, how many cars are on the Tesla lots sitting there on sold when the earnings call comes or whatever that you can just happen to Google Earth kind of like, it's sort of we live in an era of like, yeah, it's like, it's all like,

all digital information is going everywhere, right? It's going to be open like a digital digital information is like a free from people are programmed, reprogramming Adobe or making Sonic the hedgehog and in Halo or whatever, because bits and bytes can just be kind of taught to do whatever now, which is pretty cool. My ultimate techno optimist point is that humans are inherently not bureaucratic and like a lot of the challenges that we have in life are with bureaucracies that

have a lot of like waste and don't do what they're trying to do and actually get worse the more money you put into them. And I feel like machines are inherently bureaucratic and if we can figure out a way to get the machines to do more of the bureaucracy that we fuck up, that it will be good for humans in the long run. So I'm going to believe that. No, I swear all the time on this, it's where a boost strike on the all killer, no filler podcast. Yeah, that's one point. Okay, my

last word is that over the last six months, I was building my own like sort of operating software in the simplest way that I'm not a software engineer, but I built out our scaling growth system process into a set of skills that are like an, you know, operating system that walks you through how to do strategy at pilot house as we defined it. And what was very interesting to me was that I did that and then I had it run it by itself several times and had it test to see if it would get

to the same conclusion while deleting all of the input that I had put, I had worked with it crafting along the way. Where I found was that it didn't get anywhere close. It was able to make persuasive arguments that were technically wrong but like to an outsider wouldn't be able to spot. And the missing gap was the cumulative judgment that you would end and the conclusions you needed to get to by crafting along the way. And so we will, it's kind of interesting, right? Because like

as an agency, you want to scale the part of scaling that you would think would be like, okay, let's just make software that does strategy and does it scale and whatever. That's just at the

“beginning part. And so it was a very interesting experience for me. The judgment was really important.”

The second part is everybody has a plan until they get punched in the face. And like you got to

be able to carry as an agency at pilot house. The strategists don't write a strategy and then ship it off. They call that Siegel Siegel strategy, where the agency they just give this strategy. And then it hits reality and there's no backup plan because the strategists in the room to see the results. And so we're there. We have to walk, we have to build the systems and watch that thing, see if it works and see if we can scale it out and hit every real world person, like person,

system, problem, and solve it along the way to see if that's going to work. It was a very interesting, humbling experience to do that along the way with AI. So we have like, we have a scaling growth system software. We build scaling systems at scale brands. But you do not operate them

You've got to be there to figure out that there is a JavaScript code on their back and their

“Shopify that's like messing with the registration and of AOV when they do their shipping and fulfillment.”

But that's all along the way. So it's been a very interesting experience. It scales that judgment.

But again, it's the rubber meets the road. We are not in a fully digitized service here.

And I don't think that part can be removed and replaced by an agent.

“Beautiful. I think let's leave it there for today, guys. If you are listening to this podcast,”

and you made it all the way through to the end. I want you to email me at [email protected]. And I will send you one of these ebit field flannel blue hats.

“And I think this was an interesting conversation. I think it was abstract time.”

But I do think there was a lot of interesting music, a little different for the DDC podcast. But we'll have to get you guys on the run-down sometime too. And we can go really separate areas. Thanks, guys. This was a lot of fun. Yeah. Thanks, Eric. Have a great black Friday. We'll talk to you again soon. Thanks so much for listening to today's episode. If you're not a subscriber to our newsletter, you can do that right now at direct to consumeralloneword.co.

I'm Eric Dick, and this has been the DDC podcast. We'll see you next time.