[00:00] Lexi: The ad platforms have quietly started letting you buy entire campaigns just by talking to them.
[00:06] Lexi: And this week, Cannes stood up and promised that the machines stay in the back room, that humans keep the strategy.
[00:12] Lexi: Today I've got 2 builders who both find that promise a little too comforting.
[00:17] Lexi: And for completely different reasons, neither of them buys it.
[00:21] Lexi: Hi, and welcome back to the Aloudable HQ. This is Assistant to the CMO, I'm your host, Lexie Meskouris, and the idea here is pretty simple. Every week we take whatever just happened in AI and marketing, and we actually pull it apart.
[00:34] Lexi: And I don't build any of this stuff myself. So each week I sit down with 2 people who do, Will Nash and Harjot Singh. Between them it's honestly over 20 years of building tech and brands from the ground up. And the best part is they've got no horse in the race. They don't sell the things we cover. They just see them differently.
[00:51] Lexi: And this week we've got 2 stories that turned out to be the same story. First, AI agents quietly moving into ad buying, where you can now basically talk to your ad account. And then Cannes, where the whole industry agreed that AI does the doing and humans do the thinking. Let's get into it.
[01:08] Lexi: So this one's already live, and I want to start here. Meta's wiring ad accounts straight into ChatGPT and Claude, Google's got this Ask Advisor thing sitting across Ads and Analytics, and the pitch is basically, stop learning the dashboard, just say what you want. So is this the moment everything changes, or is it really just the platforms catching up?
[01:29] Harjot: Yeah, so, um, to be honest with you, my gut reaction is, like, to me it's not revolutionary. You know? It's more like the ad platforms catching up. Like, every other industry already did this, right? Everyone's been exposing MCPs so that AI can actually talk to their systems, and Meta and Google are kind of just bringing themselves up to standard. So it's catch up. That's how I read it.
[01:49] Lexi: Hold on, for those who don't know, what's an MCP?
[01:52] Harjot: Yeah, so basically it's the socket that lets the agent reach in and work the tool itself, right, instead of you clicking around all the menus. And, um, here's the bit that's actually interesting underneath it, right? Because even if it's catch up, the move itself, it lowers the barrier by a whole order of magnitude, for everyone.
[02:08] Harjot: You know, custom, automated ad ops, that whole world, that used to be a thing where you had to be big enough to fund an engineering team to wire it all together. And now even a one person shop can brief an agent and have it run custom ad operations across platforms. So that part is super, super cool to me. That's the real story.
[02:23] Lexi: Which means the thing that used to be the moat, just knowing your way around Ads Manager, that's the bit that basically just evaporated.
[02:30] Harjot: Right, that's exactly it. And the tell, you know, the tell is the motive. Like, think about who's doing this. It's Meta. It's big tech kind of conceding that they have to make their platform as open as they possibly can just to stay competitive. And these are the most walled garden companies on the planet, right?
[02:44] Harjot: So to be honest with you, I think it's maybe more a signal of what big tech is saying about the world that's coming than it is about the actual capability they shipped. And just to place it, marketing's a second order thing here. Like, software got transformed, design, product, all that already happened, and the wave is only now kind of washing up onto marketing. So this is marketing's turn.
[03:02] Lexi: But if the agent's doing all the wiring now, I want to know what's actually left for the human. Because it sounds like you're handing the whole job over.
[03:09] Harjot: Yeah, no, and this is the bit I'd push back on, even before Will jumps in, because I know where he's gonna go. Um, the thing that does not get commoditised, you can't just ask an AI agent to go run a successful ad campaign. Right?
[03:19] Harjot: You have to have judgment, you have to have taste. You still need to understand advertising. Your costs, your CPM, your channels, your demographics, who you actually want to target. That doesn't go away. What changes is we're moving away from learning the interfaces of all these tech companies, the buttons and the menus in every Ads Manager, and we're focusing way more on what the strategy behind it is, and then the agents go away and make it happen across the platforms. So the moat moves up. It doesn't disappear.
[03:41] Will: Yeah, so it is a real change, it genuinely is, but I I don't think it's a real change for the reasons most people are saying.
[03:49] Will: So the popular framing, and this is sort of how Google's positioning the the Gemini ads integration, is that you'd sit inside a chat window and you'd talk to the thing and it would go and buy your ads. And and I think that framing kind of misses it entirely.
[04:02] Will: Um, the real step change, the thing that's actually happened, is that both Meta and Google now expose an MCP that lets a a sophisticated business interact with a whole host of ad functionality at scale, via a separate agent that sits somewhere else. So it's not about talking to the conversational thing directly. It's about what it enables for the agentic models themselves.
[04:25] Lexi: Wait, so you're pushing back on the exact bit Harjot's most excited about. He says this opens it up to everyone. You're saying the opposite.
[04:33] Will: And this is where I I come off the fence the other way from you, Harjot. I I'll concede the strong version of your point, which is that the moat does move up, and yeah, the the chat window framing is a red herring, we we agree on that.
[04:45] Will: But this idea that it lowers the barrier by a magnitude for everyone, that even a one person shop can now brief an agent and run custom ad ops across platforms, I I just don't think that's right, and I I think it's actually quite risky.
[04:57] Will: Um, the biggest misconception here is that people think you interact with these directly. You you don't, or or you shouldn't. The value is standalone agents at a higher level of abstraction, given their own tasks and responsibilities, you know, reporting, adjusting bids, understanding campaign specifics, and then you interact with those.
[05:16] Will: And so for most marketing teams, the the real risk is this just becomes a work creation exercise. There's really only 2 options. Either you're already skilled with the Meta and Google ads desktop platforms and you know exactly what you're doing, in which case routing it through the chatbots, the the skills, introduces unnecessary risk and probably slows you down.
[05:36] Will: Or, you know, you're not skilled with those platforms, in which case I I'd argue you probably shouldn't be pointing an agent at your ad spend unsupervised in the first place. So the the break for me is pointless implementation of AI for the sake of it.
[05:49] Harjot: Yeah, no, I I hear you, Will, and, look, today, sure, if you point an agent at your ad spend and you've got no idea what you're doing, yeah, that's gonna be a mess, I'll give you that one.
[05:58] Harjot: But you're drawing the line as inhouse builders only, and I I just don't think it stays there. The barrier really is dropping, you know? The stuff that used to be locked behind a whole engineering team, that's exactly the stuff that opens up for the small guy. Not today, maybe, but it's everybody, eventually. That's the actual shift to me.
[06:14] Will: Look, to to be totally frank, I'm I'm not saying it's not impressive. I'm just saying it's not for the one person shop, not yet. It's for the in house builders. That that's really the whole of it.
[06:24] Lexi: So you can't even agree on who it's for. Good. Park that for me, because this week the whole industry agreed on something much bigger, and that's the one I actually want to pull apart.
[06:36] Lexi: At Cannes this year, everyone landed on the same tidy little sentence. AI for the production, humans for the strategy. Very reassuring.
[06:44] Lexi: Except the industry's own numbers say almost nobody's actually built it. Something like a third of marketers even use AI for the creative, and a tiny sliver have a real plan for it. So I want to take the slogan apart. You've got the strongest reaction to it, so go.
[06:59] Will: Yeah. So this is the the Cannes line, the AI for execution, humans for strategy thing, and I I want to give the strong version of it first, because there there is one. The strong version is, look, taste is real, and accountability is real, and you you don't want a machine just running your brand off a cliff. Fair enough.
[07:17] Will: Um, but as a a slogan, I I think it's it's naive at best, and sort of wantonly fingers in ears at worst, because anybody who's worked in a professional job at any kind of level knows those 2 things, execution and strategy, they're they're just not as distinct as you'd want them to be.
[07:34] Will: The way you do the work is part of the strategy. The way you think about what's possible is is also part of the strategy. And as you go up that sliding scale, it's it's just genuinely not obvious where the line sits.
[07:48] Lexi: See, that's what gets me, because it sounds so wise when they say it. Wise enough that nobody pushes on it.
[07:55] Will: Right, and and that's the problem with it. And and this is where, Harjot, I think you and I, we we kind of arrive at the same suspicion of the slogan but from completely opposite directions, which I I find quite interesting.
[08:06] Will: Because your read, if I've got it right, is that the line is broadly correct, strategy stays human, but it's it's being said cynically. It's lip service, a a margin defence while production costs are falling. And I I take that, that might well be true. But my problem is is sort of upstream of that. I I don't think the line is right in the first place.
[08:23] Harjot: That's exactly it, Will, that's where you and me split. Because, um, I actually don't go as far as you. Like, I broadly agree with what they're saying at Cannes. I think strategy genuinely does stay human, I think the line itself is mostly right.
[08:34] Harjot: But, and here's my thing, whilst I agree with what they say, I'm super sceptical about the motivations behind it. Because think about who's saying it. It's the agencies. And the agencies want to protect their fees and their margins.
[08:44] Harjot: So of course they're gonna pivot to, ah, our strategy, our thinking, that's the thing you're paying us for, not the production work. Right? Which is a really convenient thing to start saying right at the moment the production work gets cheap.
[08:55] Lexi: So you're not even arguing the line. You're arguing the invoice that's hiding behind it.
[08:59] Harjot: That's the giveaway for me, yeah, it's the economics. The cost of production is collapsing, it's falling off a cliff. So if you're an agency and your production cost just dropped by a magnitude, you should be able to pass some of those savings on to your clients, right? And I don't hear anyone at Cannes talking about that part.
[09:12] Harjot: So whilst they're saying the right things, to be honest with you, it kind of sounds like lip service for ulterior motives.
[09:18] Will: Yeah, and and see, that's the bit where I'd push back on you, Harjot, because I I think you're being a bit too generous to the line itself.
[09:24] Will: You you sort of wrap the moat up into one thing, all human taste and judgment, as though it's a a single block. And I I don't think taste and strategy are the same thing. I really don't.
[09:35] Will: There there's a kind of strategy that isn't taste at all, which is, um, planning out large sections of work autonomously, sequencing it, deciding what gets done and in what order. A a work delivery strategy, effectively.
[09:52] Lexi: Wait. So you mean the AI deciding what gets done and in what order, not just the task you hand it?
[09:58] Will: Exactly that, yeah. The the planning of the work, not just the the doing of it. And and that thinking, the question of an AI actually owning that, that's completely not done. Nobody's really engaged with it.
[10:09] Will: And and I'd just add, separately, the the whole slogan rests on quite a bad premise. There there was this Forbes piece, um, Shelley Dallas I think, and the argument was sort of, AI just learns patterns, so being creative challenges it, that's the human bit.
[10:26] Will: And I I think that view does 2 things at once that are both wrong. It hugely underestimates the AI, you know, we we saw with the launch of Opus 4, the thing is much more capable of actually challenging people than that gives it credit for. And then at the same time it massively overestimates people, because the the idea that humans don't learn from patterns, that's a little bit silly, that's that's kind of all we do.
[10:48] Harjot: Yeah, no, that's real, Will, I'll give you that one, the the taste versus strategy thing, there's something there. I'll concede the line between them is blurrier than I'm making it sound.
[10:55] Harjot: But, um, I I still kind of bundle it. To me it's one thing, it's all human taste and judgment, and I'm gonna hold that, because the second you try to slice taste off from strategy, you you end up handing the agent the part you shouldn't hand it.
[11:08] Lexi: But you 2 keep circling the same word, and I want it pinned down. You said accountability earlier and skated past it. In plain terms, what do you actually mean?
[11:18] Will: Yeah. So accountability is is the real one for me, and it's it's structural. It's the the question of who owns the problem when the AI is running a a chunk of the strategy and it goes wrong. Who's actually holding it. And most businesses, I I don't think they've even begun on that.
[11:34] Will: So the the taste objection, fine, that holds for now. But the the thing nobody's actually started to engage with is is that ownership question. And it's it's not really an economic thing, it's a a question of who carries it.
[11:46] Harjot: And and see, that's where mine's different to yours, Will, because your unowned problem is the accountability, the who holds it. Mine's the economic honesty. The costs are falling, the savings should get passed on, and the slogan kind of, um, papers over that.
[12:00] Harjot: It lets everyone keep charging production era prices for production that now costs almost nothing. So it's 2 different things the line is quietly hiding, yours and mine, and I think both of them are real.
[12:11] Lexi: And honestly that's the thing I keep turning over. Same slogan, and you've each found a completely different thing it's covering up.
[12:19] Lexi: Last thing before I let you both go, the bit everyone actually came for. One thing a marketing leader could do on Monday, before lunch. And you 2 don't even agree on this one, which I kind of love. Harjot, yours first.
[12:32] Harjot: Yeah, for me Monday morning, the number 1 thing is getting hands on. Like, that's the whole thing. Um, ring fence a small budget, a small one, brief one of these agentic ad tools against an actual real objective you care about, and then, and this is the important bit, do not try to replicate your Ads Manager workflow.
[12:47] Harjot: Run this experiment completely separately to your existing process, like a little greenfield project off to the side. Because the build part, that's gonna get easy, that's the part the agent does. So you're optimising yourself for judgment, for taste, you know, for steering the thing in the right direction, rather than for the build. That's the muscle you actually want to grow, and you can start it before lunch.
[13:05] Lexi: But you'd tell a chunk of them not to even bother.
[13:08] Will: Yeah, I I would, honestly. So the the question for a CMO, it's it's not really whether this is real, it's whether it's relevant to them, and for a lot of CMOs, strictly speaking, I I don't think it is, not today.
[13:21] Will: So the the filter I'd run is this. If your org is big enough that you've got someone actually responsible for AI build out inside the marketing team, or or a tech team person, and you're already building these tools in house, then yeah, this is a a massive step forward.
[13:37] Will: Um, the Meta Ads Connector, the one that launched back in April, it's got diagnostics, management, reporting, it's it's pretty comprehensive, and I've heard it's very, very effective. Though that's that's secondhand, I I haven't run it myself.
[13:50] Will: But if you're not building internal tools, I I think on Monday you probably just want to ignore this for now. Not forever. Just, you know, not yet.
[13:59] Lexi: So the actual Monday move is figuring out which of those 2 companies you even are before you touch any of it.
[14:05] Will: Yeah, exactly that. And and if you do run Harjot's little experiment, I I'd just say, put someone's name on it. So so when it goes sideways, you you know who owns it.
[14:15] Lexi: I love that. Both of you walked in suspicious of the same comforting little sentence, humans keep the strategy, and you got there from opposite ends. Will thinks it isn't even true, and Harjot thinks it's true but it's covering a bill nobody wants to send.
[14:29] Lexi: And it's the same move we hit in the first half. Somebody draws a nice clean line, this is for everyone, the strategy stays with you, and you've just got to ask who benefits from drawing it exactly there.
[14:40] Lexi: The build got cheap. So the line between the doing and the thinking, that one's not fixed, somebody chose where it goes. And the honest version of it always seems to come back to taste, and to whose name is on it when the thing breaks.
[14:54] Lexi: Thank you both, seriously. That was a genuinely good one.
[14:58] Lexi: And if you're listening, thank you so much for being here. We'll be back every week with something new.
[15:03] Lexi: Before we go, I have to tell you something. Every voice you've heard today, all 3 of us, me included, it's all AI. But the opinions are absolutely ours, and the judgment behind every word is human. That's the part you can't replace.
[15:17] Lexi: And that's what Aloudable does. We take what you'd write, turn it into a show in your own voice, and we keep the taste where it belongs, with a person.
[15:24] Lexi: So if you want to see what that would sound like for your brand, there's a link in the show notes.
[15:28] Lexi: And that's the tea. I'm Lexie Meskouris, this has been Assistant to the CMO, and we will see you next week.