He Creates 1,000+ Ads Every Month With AI (But He's Still Skeptical About It). transcript

You must watch or listen to this episode if you are in meta ads creative strategy anywhere at all. If you have seen the influx of AI tools hitting meta ads, you want to adopt them, but you don't know where to start. You feel overwhelmed. You see overpromising and underdelivering in all kinds of places. Man, this was such a helpful conversation for me from the best AI first creative thinker I know of in our space in Alex Cooper.

Alex runs ad crate which he's built with a few folks and he is a brilliant thinker because he's extremely AI forward integrating AI into all kinds of stuff he does but is also really AI skeptical about the impact it can have on the business and the result is a sober but advanced perspective an envelope pushing perspective on how to use AI in your ad creative approach. I was helped deeply by this episode. I've been helped by his content for a while.

You're going to love this. Listen to Alex Cooper. He's extremely helpful. Let's get into it right now. Alex Cooper, nobody I know is blending the two sides of being extremely AI forward and at the same time I would say almost AI skeptical. That's probably too strong of a statement. You can frame that however you want. um for ad creative for meta ads and and other digital advertising channels than you you have a really interesting mix of Yeah.

Yeah. like some of the most aggressively forward AI creative content I've seen and then at the same time you're tweeting things like ugly ads are ultimately what's going to win here. Um so that's the startup for this conversation. is why I was excited to talk to you is to talk about sort of what um how we blend those things. React to that comment. How do you think about what I just said? Uh look, I know it sounds confusing.

Uh I am very aware of that and I think that's a pretty spot-on explanation. Like I am very uh aggressively AI forward. I am also incredibly skeptical. Um, I think that, you know, a few months ago when when I would tweet out things like, you know, I think it's very possible in a few in like less than two years time that entire DTOC brands could be run by a handful of people and an army of AI agents. People thought that I was being extremists.

People thought that I was doing it for engagement. I think that, you know, for the first four months of this year, people started to realize that that is a very real possibility and a lot sooner than a lot of people think. At the same time, um, we're all now very aware that the feed is going to be flooded with all of these AI ads, AI statics, AI video ads when someone finally cracks like how to interact with a product on video, the same way that um, Chachi, he just did with the update.

Uh, in a feed of AI ads, how do you stand out? How do you signal to people that you're not just another like soulless company uh creating uh AI ads on the timeline? And I really do think leaning into things that AI can't replicate or at least can't replicate today is a very smart move for a lot of creative strategist agencies and brands. Doing things like you know like founder ads like you know ads from inside the warehouse ads with multiple people um where they're getting like genuine authentic reactions or just single person like authentic reactions.

It is very difficult for me to see at least in the short to midterm AI being able to replicate a good creators genuine authentic reactions. Like I'm I mean I think it's already at the point now where it can make as good if not better mashups with existing content than humans can. I think it won't be long till it gets to the point where it can make as good a like single person testimonials. So I'm holding this water bottle and I'm talking about this water bottle.

But like what about if it's me and you, you know, I'm I'm saying like, "Hey, I just bought my friend like this uh piece of licorice and I'm getting him to react to it for the first time as a gift." And like your actual reaction like maybe AI gets to the point where it can recreate that and like the the tone and the inflections and like the authenticity maybe. I don't see that being anytime soon. So I really like I I really do see the value in both.

I think there's going to be value in AI because it will allow us to do volume and it will allow us to get learnings way quicker than we otherwise would have done if we're making them value if we're making them um like manually. But at the same time like the stuff that AI can't do or can't do yet is going to be where humans can add their value by making ugly authentic ads. Uh so just the combination of both of those is what we're trying to integrate into our creative strategy as uh as more of this tech comes out.

Let me run a framework by you because I agree with everything you just said. So let me let me run a framework by you and you tell me what you think of this. Okay. Um, what if what brands do or agencies do or whatever is you build a testing process that is as aggressively AI forward as possible and you then um build so we have this framework explore and expand right our idea is like this is not this is not rocket science but it's helped us to think through what we're trying to accomplish at the level of creative I think you've said some similar things times right explore is the way you test new angles new messages expand is you have a message or an angle that works and And then you try to expand that message and angle by taking the same message and then retelling the same story or restating the same message in as many ways as possible as creatively as possible.

There's obviously a place for AI in that. But what I've started to think is what if the actual way to use AI is you use the AI primarily on the explore side. And so the idea is as quickly and as cheaply as possible you test messages. You use explainer videos. Because we've got I've got one brand using AI avatars and has spent hundreds of thousands of dollars on AI avatar ads that work. They I think they are not as good as like a good creator like you said as a good actor something but they are good enough and they and then often you know they're they're they're all using 11 Labs for VO. um you know the script is heavily is a mix of AI and human the video editing itself um as a mix of AI and human you know so so there's AI all throughout the process including the actual avatar on screen green screen style ads that kind of thing so um that has allowed us to test a whole bunch of messages at scale really fast and really really ridiculously cheap.

Um, but what we are also finding is that then maybe to to then take that message that works and and then hammer it home with a bunch of different styles. Again, there's some place for AI in that and we've certainly used it, but increasingly what I would do is say, okay, what happens if we say we want to get creators to go read these scripts or get creators to go to give them a pro a brief and have them do stuff? Send influencers packages with briefs about what kind of content we're looking for. start doing that with stills, start doing it with, you know, whatever.

Like just all of these different things. Um finding different ways to make this same point with um with real people basically. And essentially that is the place where AI has the biggest impact at this point in the game, which is the ability to very cheaply test a lot of message ideas in inherent in that is my assumption. And we've never talked about this, so tell me if you disagree with this. We totally can disagree.

You don't have to sit here and be like, "Oh, you're so smart. Great." You know, um uh you're welcome to disagree with me. Um, inherent in that is the assumption that messaging is what moves the needle first and foremost, right? That like that the the key to great advertising is the message. Um, and so so what you're trying to test in that testing phase is not formats, it's not creators, it's not anything like that. It's the message.

Um, and so when you have a message, then you have to expand the message and you want to then make that message work as effectively as possible by scaling it up with better quality, you know, creators, etc. What do you think about that way of approaching this? Is that too narrow? Yeah, I know. I think I I've got a lot to say here. I I I think you're spot on there in in I think you're spot on there in in in the like the biggest impact for AI is in the exploration phase right now.

I think eventually it'll be able to do both at a very high level. But like I mean we don't need to like talk about here like the importance of of like creative volume and how important it is to to get a lot of ads inside the ad account and test different messages uh and different angles. I was uh I was in a conversation with um uh with someone recently like it's a it's a brand that we used to work with and uh they um they just went for a pure volume approach and I'm I'm not saying that like quality isn't creative quality is not important because it absolutely is but to your point of like the AI ads that you were generating in like Arcads or like whatever the platform was the AI video ads you were generating you look at them objectively they're not they don't look like wow this is this is a great at uh you would think that a human can make something better, but it's just the volume that you're able to produce them at uh means that it's such a a massive value ad for brands and and this specific brand like if you looked at their ads versus their ad library versus their competitor's ad library.

Um you would think that they're at the same level or uh or like worse than their competitors because their ads objectively don't look that good, but they just tested so much volume. I mean, this was before like all of these tools came out. So, they would they just had like really good creative operations, but now it' be 10x with with with the air tools. They just did so much volume that they were able to learn way quicker what messages worked, what messages didn't work.

Um, so that like I mean that she ended up growing their business like from 10 to 50 mil and and like literally credited uh the amount of creative volume and message testing as one of the key contributors towards being able to scale uh that quickly. So like I I definitely agree that exploration uh like in its present state uh AI is the is the biggest value ad and like to to your other point as well about like the messaging being the most important like if if we're in this position where we're not we're saying like okay creative agencies aren't going to get fired like you know creative aren't going to get fired by AI it's going to enhance them and it's like right so how do you make how do you make high performance creative today with AI and like like you said people want the answer that's like oh you got to try this format or you got to try this style or like you got to do this but honestly the answer doesn't change with the development of AI it's still you need to have a deep understanding of psychology you need to go out you need to go and read on advertising you need to read Martin's mind stage you need to study the Harman brothers you need to get an understanding of psychology because then with psychology you become better at research and then become with better research you become better at copyrightiting and like that's where the headlines come from that's where those come from that's where the the scripts come from and like I I I just think that making high performance creative is about like not necessarily what format of style you choose but more so that like okay I run an anti-dandruff shampoo brand and I saw in one uh comment on a Reddit thread deep into the like deep into the depths of Reddit, I saw someone said, "Oh, actually like I was really struggling with beard dandruff and like this actually didn't this didn't help my hair.

This helped my beard." And like you test that as an ad and like it becomes your best performer because you've uncovered this secret painoint that only insiders who have that actual pain point would understand and relate to. Like that's what I think being a good creative strategist is. And I don't think that changes with the I mean the other one the other one that is probably more familiar to people is like the I'm sure the fa killer example that Ash posted from Obby like where fer is an inside term that only people who have that painoint understand and they did that as a static just faill and it became one of the top performing statics because like you know they understood the words and phrases and use cases that their customers actually have of their product.

Um I think the only thing that changes um with AI is now that you just have to understand how to do all of that plus you have to understand how to reverse engineer that output with AI. Like you just have to do psychology, research and copyrighting which are all kind of into spine anyway plus prompting which is understand how to take all of this and reverse engineer the output with AI. I love Intelligjam. I love them because my clients love Intelligjam.

They've been using them for a long time. It is the CRO software of record in our space. And that's because if you aren't thinking about your business with uh attempts to build real operational excellence all the way through your business uh where you're thinking profit first. How do I build a system for deploying tests that lead to generating profit in my business? Intelligence is an incredible tool for that. I have used Intelligjee in the last couple of weeks alone to test the price of a product which is amazing right like product price has a massive impact on your profitability.

The whole tariff thing that has come up recently is one of those things where like you might have more or less price elasticity than you think. You won't know unless you test it. And that's a really big deal if your margin gets squeezed all of a sudden. And even if your margin doesn't get squeezed, it's a really big deal because there may be more margin for you. You won't know unless you test it. So I have a client who is testing one of their core products and testing the price of that product at multiple levels.

You can do that with Intelligjems. You can al I've also tested in the last couple weeks like traditional across the board CRO test just like an AB design type test to see if different elements in the page would impact things differently. You can also test free shipping thresholds. Sitewide offers is that 10% off offer you're offering new customers when they sign up for your email list actually helping or is it just cannibalizing margin from customers who are going to buy anyway?

You can answer that question definitively. Your free shipping threshold, you can answer that question definitively by using Intelliggeems to do it. Intelligence is all about optimizing your store for profit. Not just conversion rate, not just AOV, but profit. They actually tie in to your unit economic numbers in your Shopify, like the actual cogs on your products and then measure everything by profit per visit, which is their golden key god metric in Intellig.

And they spit out everything there. They tell you if you're a stat sig on your tests. They give you confidence intervals. It's really, really great software. It's easy to install. Just stop right now, listen to what I'm saying. Go to intelliggeems.io. Use the code ferris 20 fs 20 to get 20% off your first three months with intelligence today. It's great software. You should be using it if you're serious about building a a good business in e-commerce.

Intelligence.io. I have spent a lot of time on calls recently looking at AI tools and and things like that. And I'm starting to believe that it's actually been kind of a waste of my time. Uh well, it's probably not a waste of my time, but um what I'm starting to believe is that the actual way to do this is to get clear on my first principles like you're saying about what makes good ads, to get clear on my process for making ads right now, and then from there to turn to find ways in which AI can speed up and um make make that process that already exists more efficient as opposed to the promise of some AI tools right now, which is like um we'll do it all for you.

We'll do all of it. It'll just be press a button and it's the infinite ads machine. And and especially on the idea that volume is like the only thing that matters. It may be the case that you can produce so much volume so cheap and test it so cheap with manual bids that you don't actually need even the step that I said about first principles. like you might just be able to take so many freaking swings that like it just works.

Um I just I haven't quite seen that yet. Um instead where I've seen it work the best is taking the things I already believe and then improving the supply chain elements of that process better. So you just you just gave one of these which is you went to Reddit and found a comment down in the down in the deep dark part of the internet called Reddit into the depths of the belly of the beast of Reddit where you you should never make a comment because you will you will get eaten alive by the crazy Redditors.

Um, and I saw you I see you tweeting something about this too, but like um what that reflects is actually just a principle which is that you need to understand how your customers talk like you just said and understand where their actual pain points are and now you need to um go learn that and then figure out how to turn that part into an AI um enhanced process. Right? So AI just takes some of the drudgery out of that work for you.

Actually, this was my first concern about AI was that it in fact what it would do would it would remove too much of the mental churning that I do in a process like that and therefore it would make give me less creative insight for the things that are really going to be needle movers because I actually in some ways want the friction. I want my mental gears to to churn because that is that is that is the process of human problem solving.

Um and arguably AI can just do that better than me so I shouldn't want that, right? Um, but but I' I've sort of worried about that. Now, what I'm interested in is like if you have taken that Reddit process that you just mentioned, discover the customer painoint, and found an AI way to to remove some of the drudgery of that part of the process, speed it up, and then and then take the basic idea, which is know how your customers talk and then talk the same way and promise them the things that they want.

Take that same idea and and improve the process that you already have. You already have the process. Now, you're just trying to document it and improve it with AI. Is that what you're doing? Yeah, I I think that's definitely a much more defensible um approach to take than the former. Uh I mean like the problem with a lot of these AI tools that you're seeing at the moment and there's and there's only going to be like I mean since the icon launch I think everyone's just tried to build an AI creative strategist or an AI tool because they've seen how well that went.

Uh the problem is that all of these tools are mainly built by SAS companies or developers and like the execution here is not the moat. Uh the moat is in the understanding of what we spoke about earlier is it's like how to make an actual good direct response ad. It's really hard to be a creative strategist and like the execution like we've seen with websites like you know lovable came out in what was it two months ago now and like now anyone can uh can build a website like I may be wrong and it may be like you said that that with the amount of volume it can do like it just knows um what works and what doesn't and AI can can produce good quality ads.

Uh but the execution becomes commoditized very very quickly. I mean, we're working on on on a couple of things with integrations with like, you know, Reddit, YouTube Shorts, Pinterest, Tik Tok, but they're going to be everywhere. Like, they're not today, but like I really don't think it's long before everyone's got the tool where you can hook into every single platform that you use. And then it's just like the moat is how good at you actually at creative strategy and then how good you are reverse engineering that process through the quality of uh of your prompt.

Then to take it one step further like then you can get into MTPs which is where like instead of you connecting directly like to the APIs of a platform like the AI comes to you and p like meets you where you're at. So like the AI is in your Slack channel and the AI agent is in your creative team's brainstorm and then instead of like you know you having to run the task every time saying like go and scrape Reddit and um uh and find me all the things that people are saying that we could use or like find me all the ad comments that people are uh uh that have occurred in the last seven days that could be turned into com response ads or whatever it is.

The agent will be proactive and say hey like I heard you guys talking about this in your creative brainstorm last week. I've actually gone ahead and brainstormed uh three scripts off the back of those ideas. What do you think? Um so I mean yeah the problem is like it's hard like for me to today cuz none of these tools exist. I I really think we're talking about a matter of like a a handful of months instead of years which most people think uh these tools are away but on like as of recording today like just a lot of the tools out there I think are built by people who aren't in the space or aren't in the space like actively in the trenches building ads that convert.

And that's why there's a bit of a disconnect between like, you know, all these tools that do very well online and they all look very flashy, but there's actually no ongoing utility to the creative strategist who needs to put out 100 ads a week or like whatever that number is for you. Yeah. My experience with the tools is that they are often like 60 to 70% good or something like that. Like if there's a 10-step process, they do five of the steps awesome.

But the places where they mess up um on the other five are too big to use it uh to be an all-in-one thing. And in fact, if you build in that platform, it also becomes so clunky that trying to then manually as a human step into the pro platform and then go go fix the problems that the platform can't do all the way. Uh that that process now by having to reintegrate the human into a new process actually takes longer than if you had just done it yourself even though it could do five other things really well.

That's my experience right now. So that so instead of doing the the sort of one-sizefits-all um thing, this is why I come back to the idea that like what you actually need to do is kind of what you said earlier. Have a process. Use AI tools where possible in your existing process to speed up your already good process. Okay. So I'm I haven't I haven't fired a single person yet for my teams. Um all I'm thinking about in fact what I'm doing is like trying to pull them more into more of these conversations. like my creative team in the Philippines, things like that.

They're really smart, really talented people. So, I try to pull them in and say um and say, "Okay, how do I then um give you tools to take the current process you already have and make it better if they if they can do that?" And what I believe is that what you said is right, which is that um right now a lot of people are building tools on top of the existing tools, right? And that's great and fast and whatever, and sometimes there's a place for that, but I'm kind of like I actually think I'll probably spend too much time trying to go find which ones and where to put those if I do that too much.

It's actually probably a better use of my time to just wait the three months for the the tools themselves. Like if you're in Canva for Canvas AI tools to get so good basically because you know this is what they're all going to do, right? The Slack thing you just mentioned is for sure going to happen at some point. Meta Ads Manager is going to have like a a comment reader for you that's built into the tool where it's like you know hey some you know I don't know.

So um so increasingly all the tools going to be built AI first. What the actual comp in my mind to all of this is the advent of the internet. Uh, how old are you, Alex? I'm I'm 26 now. So, I was very Okay. Yeah. Yeah. Yeah. Yeah. Yeah. So, I'm 41. Um, I remember the first time I went into a computer lab in my junior in my junior high school, you know, the computer computer room, and it was like, here's Netscape Navigator.

I built Geio Cities websites and things that you probably don't even know what they are. And I um and and it went from this thing that I had to go to a computer lab to use to where it's just everywhere all the time all around us and it's now a normal part of everything we do. Um I think that's I think that's basically happening at the same time and it's going to happen much faster because AI can self-replicate really fast, right?

Like you can AI can use AI to go fast. So, um, so anyway, that's that's kind of the way I'm thinking about it right now is like that's is is like the sort of the level of investment I'm putting into AI is like keep being open to what's happening. Try not to be like overwhelmed, but like don't overhaul my whole processes. Instead, do the things that I know work because actually those things were already scaling brands plenty fast.

In fact, in fact, brands can't sometimes scale a whole lot faster because it's actually too financially difficult um and for a lot of reasons. So, um, so yeah, that's basically the way I'm thinking about it. And and yeah. Yeah. I mean, I I saw a meme the other day, um, that was like, do you know the the Drake meme? The hotline bling Drake meme where it's like, um, it's like uh the caption was like on the bit where he's like going like that where he's like saying no.

It was like um write an ad script yourself and then like the bit under where I was like yes was like spend 30 minutes in clawude like you know prompt engineering a script and and it's so true like we all think these tools are going to save us time and it's like actually when you factor in how much you know this doesn't do the whole job this does maybe like the middle 60 or 70% and I've still got to do the pre-work before and the after work uh actually just would have been better off just doing it myself.

Um I I do think that it's it's a smart position to take in in in that just you just kind of sit back and and let the technology develop for the next few months. What I will say in the meantime, you really can get the 8020 uh by just building up a like a prompting and context library for yourself. I really think that like more strategists should be doing this um to like increase the quality of of the outputs you get from GPT. like you said, it just goes back to like taking the process you've already got and and turning into AI rather than looking for an AI process and and trying to shoehorn that into the into the stuff that that you already do.

So just taking like the prompts that you've got and uh and like instead of having to end them any every time I just have a prompting library like whenever I want to analyze customer reviews whenever I want to uh analyze or I come up with ideation or do headlines like I've got a library I can just take and copy that straight into GPT and then when the text's there I can just insert those into my agent and like I'm always iterating on those prompts like every day.

So, it might start off with one prompt that's like, uh, here are some custom reviews. Pick out some static headlines and then I get the output and I go, actually, I want to add some more context. I want to say that these static headlines, um, must be relevant to someone who has absolutely zero context and does not care about the brand. And then I still it's still not giving me one. And then I say then I want to I want to take one step further and I go, okay, why?

So, I say, "Here is customer reviews. Give me some static headlines. Uh, make sure it's relevant to someone who does not know or care. Here's an example of a good headline for my brand, and here's why. Here's an example of a bad headline, and here's why." And like I just keep iterating on that prompt and iterating on that prompt and adding that to my prompt library uh until I've got some like really solid prompts that like near enough like every time I put custom reviews in, like they give me good ad headline outputs.

And just doing that for prompting and doing that for context as well. Like I mean what's crazy is like you can literally just go and take resources off the internet, put it into your context library and then just dump your context library into every time that you write a prompt. So like I do this all the time um inside of Poppy, which is like one of my favorite tools uh to use. I've taken like my YouTube videos like I take like I'll probably do it for this one to be fair.

I I take like an hourong video that I did with um a podcast I did with Mela or or like or Barry in the space. I put it into Poppy and say, "Hey, turn this into a a document on how to how to write static ads." And now I just use that as context when I'm making static ads in the future. I can just say, "Hey, like here's the the prompt. Go through these customer reviews. Pick out some headlines that are good for someone who has zero context.

Um here's a good example. Here's a bad example. Here's why. And by the way, here's a five-page document on how to make great static ads. And like there's no way you do that and keep iterating on it and not improve the quality of your outputs over time. So yeah, I mean you you can use a lot of these tools and and I know it's it's overwhelming because even me who's someone who like I would agree that I'm very aggressively into AI, I still feel overwhelmed.

I go into Twitter every day and it's like damn man, there's a new feature or there's a new update or there's a new launch. Uh, I just feel like since the turn of the year, it's been it's been 247. Uh, it's tough, but it's just working out like, you know, what are the 8020 like key tools that you want to be using in your workflow and then just building out your own prompting and context libraries. On top of that, if you are scaling up your creative output, you should be doing it with help from my design and video editing team in the Philippines.

It is the same team that I use, that I'm scaling my clients brands with. They're called Behind the Scenes Studio. They're an offshoot extension, sister company, whatever you want to call it, of more staffing, who you've heard me talk about for a very long time. Philippines-based team that is laser focused on graphic design and video editing for meta ads. And the beauty of that is if you like the way that I am thinking about this, if you like the way that Alex is thinking about this, they're going to be a great um extension of your team because they are thinking the same way.

In fact, behind the scenes built their creative process off the back of my creative process. We started working together on some things. they realize there's an opportunity to take the kinds of stuff we're talking about all the time and extend it out to a bunch more clients. And so if you are trying to focus on building a high volume of good quality meta ads creative at a low cost, they are the perfect solution to that.

Um because they are based in the Philippines. So um there's an opportunity for you to uh to pay a good rate in the Philippines to get great talent, but at the same time pay less than you'd pay equivalent talent in the US. Behind the scenes is um attracting really good quality design and edit talent. One of my designers h was working at like a major serious magazine in the Philippines before coming over and working with me um because the the opportunity was so good.

Like these people know what they're doing. They're really good. If you think Philippines equals cheap equals bad, you have it wrong. Um you were thinking the wrong way about this. Behind the scenes is extremely high quality in what they do. Built for meta ads first. You're going to love working with them. So go to btsstudio.co. btsstudio.co. The link is in the show notes for that. Get a call started. See if it's the right solution for you. btsstudio.co.

I highly recommend them. I'm glad to hear that you feel overwhelmed, too. Maybe I shouldn't be. Maybe I should Maybe I should It should make me more overwhelmed if you're overwhelmed. But yeah, it's good to know. What I've actually um come to with that is that they're actually part of the reason that I feel like I have to be so AI forward is that I feel like I'm going to lose out on the opportunity to make a bunch of money. um like like my clients are all going to fire me if I don't aggressively adopt things fast enough because they're all going to go to, you know, you or something, you know, whatever.

Uh uh and not that you even are really competitive to me. I'm not, you know, whatever it it's just it's just an irrational fear I have, you know. And the truth is what I think is that actually I can make my clients a lot of money right now by doing a really good job at the things that we do and like I said, using AI to support them. So I don't I don't feel the need at this stage for an early adopter mentality to overhaul everything that I'm doing.

And if I sort of am able to identify that fear a little bit, then I'm able to kind of calm down a little bit and then try to actually just figure out, okay, how do I just take the things I'm doing now and make the process better? Sort of um be slightly less optimistic about what it can do for me um than than what I'm doing now. So yeah. Um, really fast to go back to what you're saying. Um, when you say you've got that context library and those prompts, I think that's really good input.

I'm just starting to get better at doing the thing you also just said, which is like take videos of mine and turn it into um, you know, processes and documentation like all that kind of stuff. Um, so yeah, I'm still learning to have that instinct where like everything I use AI for. But um what tools do you use actually for the prompt um creation for like you know you mentioned like read these reviews here's the here's the example ad here's who you're talking to you give it sort of all that information is that a cloud project like is that as custom GPT what are you doing for that yeah well first of all I will say I love your approach of just like kind of recognizing that like it's overwhelming but we don't need to overhaul our processes just leave all the like frantic worrying to people like me um we can just we can yeah we can just we can just be up all night trying to worry like if we're going to get replaced by AI and like you just you just sit tight.

Um but I I think it's the right way to do it. Like you you shouldn't be you shouldn't be overhauling your processes because you can overhaul your processes and then in two months time you have to do it again because like we're so early to this stuff. It's like you don't know and like like AI is going to come and significantly change the way we work but like marketing principles are always going to stay the same. So, if I'm going to bet on one um to be the center of my strategy right now, I'm I I know what one I'm taking.

Um as it pertains to tools, like I said, there we I I try my best to try as many as I can and honestly like most of them are very average and look cooler than they are actually practical. Uh in terms of the ones that I actually use day-to-day, um so actually we'll start off. We we spoke about Reddit a while ago. Um I think Gigabrain is a insanely useful tool. Uh it basically acts as chat GBT but instead of the uh instead of the knowledge base being what GBT has, it's it's Reddit.

So like you can find real words and phrases that your customers are using, you can just type into the chat like you know I have an anti-dandruff shampoo brand. what what specific pain points and like language do people have uh about dandruff when they have that pain point and it will go and create Reddit. It will link the Reddit threads and also look at YouTube comments as well. That's really gigabrain you said. Gigabrain.

Yes, it's $10 a month as well. Like honestly, it's killer. Um that's good. Something slightly more expensive. Deep research uh is is incredibly powerful. I mean like I mean the the what everyone was saying at the start of the year that was like oh by the end of the year 50% of a strategist work uh on research will be done by AI. I genuinely think deep research can do 50% of your research work today. Like it's scary the amount of detail that it will go into and and produce reports or insights that would have taken hours uh to do manually uh or even in a normal chatbt subscription uh in in about 20 minutes.

So again, I'm not doing anything that I wouldn't do manually or I wouldn't do with the chat GBT pro plan. Uh I'm just asking it to go through articles, go through Reddit, uh go through YouTube comments, like all the things that it that I would do and just uh and just find the keywords and phrases like tell me what I need to know, give me ideas that I can use in my creative strategy. So deep research is great. Uh clawed projects again going back to what we were saying earlier of of context libraries. uh you know you can either use uh memory inside of GPT or you can use clawed projects.

So for each of our clients we have uh project setup so I don't have to dump you know client context in every single time but like there's context on like your brand and there's also context on like industry expertise. So, like going back to what we were saying, like it it is an instinct like going to find like this podcast, downloading it or you don't even download it. You can just put the YouTube video inside of Poppy and it will transcribe it and and and use it straight away.

Uh but just collecting a bunch of resources that are really great resources and like there's a temptation to want to put everything um into just a a GPT or into a Poppy board and say go and work on this. I prefer to pick like a a small group of people that I really believe in their philosophies and like methodologies and just take long form content from them and dump it in or sometimes I'll I'll do my own stuff because you know I I believe in the stuff I say uh and just say like here's a YouTube video like I I did one with Dar Denny recently like I took one of her recent YouTube videos I thought it was a really good video.

I put it in and I said, um, put turn this into a document on on how to, uh, test ads and then like it's then I can use that as context in the prompts or in the agents that we set up. It's it's all just about like who can build the like who's got the best documentation. Uh, because like as we said, GBT and cord are inherently not good creative strategists. Like you need to teach them what is good and what is not good.

They're great reasoning models, but they're not great at making performance corre like performance direct response ads. Um, so we need to teach what is good and what is not. And that just comes down to the quality of the context you give it. Both both on the the basis of like here's my brand and here's what you need to know about my products. And here's what actually a good output looks like. Here's how to make a static ad.

Here's how to do good research or here's like what you should be looking for. So yeah, I am a I'm a big proponent of um deep research and then Poppy for that reason. Like it's it's so useful uh because of course you can't drag in URLs or like you know many different inputs into um into Claude uh natively. But what Poppy is is is like an AI wrapper where you can choose the model you want. You can choose it can be floor 3.7 or 40 or 01 like whatever uh LLM that you want to work with and then you can really easily insert different um inputs.

So you can insert like YouTube videos, you can insert Tik Tok videos, you can insert images, you can insert your customer reviews if you want to websites, you know, review pages, you can even voice message it. Um or like like just record a voice note on like hey um here is like some context about my brand. give it examples, good and bad examples, and then just chat with Claude as if you were chatting inside of Claude, except Claude has this all of this context.

It's like Claude projects, but you've got all this additional like context from different parts of the internet that are very hard to get into Claude uh natively. You don't have to like, you know, control A and C and then control B into uh into Claude, which you've all done many times. Um, so I think Poppy really is great not just for like its ability to to make ads and and and uh like headlines or scripts because we write all of our scripts inside of there, but also just to build context libraries.

If I see a really good piece of content, whether it's a newsletter, um whether it's a tweet, whether it's a a YouTube video, like I save it to my context library. I turn it into some kind of document inside of Poppy. And like I'm just building my context library uh and building like the almost like manual on how to make a great static or how to make how to do great research or like whatever the task is inside of creative strategy so that we can feed chat GBT, we can feed Paul, we can feed our agents at some point when we need that context.

The thing that's really interesting to me about all of this is that um what it reflects is that the real challenge of this kind of work and of most kinds of work is actually the replication of process. So like you Alex already have a whole bunch of ideas in your head about what works and what doesn't. But of course they float in and out of your head. At times you forget that you believe concept A over here about something about creative strategy and over here you have concept B about your research method.

And you know, no matter how much you believe those things, you're just in your process somewhere of doing research for an ad. And some of those things are just more present to your mind at times than others. And you believed this thing that was really important six months ago that you sort of forgot about for no reason. You didn't stop believing it. You just thought of other things. You know, it's just really hard to maintain this um these things that you believe, right?

And so uh and so it's interesting how much of what you're expressing here is about the ability to take those beliefs and essentially not lose them actually to tap into those beliefs in your actual creation process. That strikes me as like a great use of what AI is because um AI doesn't forget things you know like by giving it the documentation um you know it's going to have that in the context you know library in a way that you and I do you know don't.

And then of course that's before you even talk about the step of replicating that across other people in an organization, right? Which is which is its own challenge. Um and so yeah, you know, the tools you're talking about there I think really help with that. Yeah, that's a great point. Like I you know, I can't build this context library overnight because as you say, like things flip out your mind. You might think something more strongly today than you did 3 months ago.

So it's over time like I'm I started out with what I thought I knew. Um and then I just I add tweets that I see as they come in by the day. I add I add YouTube videos. I add newsletters and over time you it just compounds and compounds. One thing that like if I what I would do if I didn't have any content out there and like I had to start from scratch and I let's say I didn't have any SOPs, I just knew what my process is and I didn't uh I didn't have anything documented is I would record a podcast with myself.

Um I would tell chat GPT uh this is what I'm trying to do. Like I'm trying to build out a context library and I have nothing right now. That's genius. Um, yeah. Yeah, I use this tool, Riverside, that we're recording in. I This is actually one of my favorite uses of AI so far. I used it I actually used it for some X posts. I'm a pretty I'm a pretty verbal processor. You may have noticed. I'm sorry to interrupt you. I'm a very pretty verbal processor.

So, what I did is I realized, wait a minute, why don't I give Claude some of my best performing tweets so that it has some sense of like what my voice is and all that stuff. Tell it who I'm talking to. Of course, I did the thing that I always try and do with AI that I'm still getting the habit of, which I asked Claude how to give Claude the right instructions. Right? So, yeah. Um, so Claude said, "Okay, here's what I need from you to accomplish the thing that you're trying to accomplish." I told that and then I said, "Okay, I'm going to give you a transcript of of me talking.

I want you to turn that into a tweet." Basically, um, and it got it about 80% there. And it allowed me to um to do Oh, no, it was a newsletter. Excuse me. It was It's not just a tweet. It was a newsletter. Um, so it was longer. It was a longer form piece of content. And I I said like this is the thing I'm trying to say and I wanted to talk through some ideas and connect some ideas out loud. So I talked I recorded it.

I had to spit it out. I gave it in my newsletters and then yeah and then now I'm I'm uh and then I edited and and you know pasted and it's like man that saved me so much time and helped me it it it it helped me create better content because in because it it sort of blended together the work of writing and talking. So as I talked I could get the ideas out and then and then in the editing process I wrote and wr writing is an act of thinking as well.

So I don't want to completely eliminate my writing process. So um so I want that to be in there. So as I was working through I was reading writing just cleaning things up and in the end I don't know that it saved me a huge amount of time but I think it made the content way better. Um and and yeah I think um it definitely saved me some time and allowed me to like get through the mental barrier of sitting down with a blank page and trying to start writing and all that kind of stuff.

So um yeah for the same thing for SOPs for sure. I I did this exact same thing with my team. I was looking at like, okay, we're trying to do something with the way we, you know, launch ads into an ad account and whatever. And I thought, I'm just going to talk and then I'm going to have um chatbt turn that talking into instructions. And it did it. In fact, it it it outputed it to me in the in a notion format, right? So, I could actually give me a notion file that I could put directly into notion.

Again, it was 80% 90% of the way there. Made a couple little edits. It was it was like a great way to do it because I I mean I I hate the work of doing the detail process, typing out each stage of things. It feels like drudgery to me, you know? So the idea that Shashi Boutique functions as my executive assistant in that case who's like doing that work for me was like incredibly helpful. Yeah. I mean spot on. I mean have you have you have you used Poppy before?

No, I didn't know about it. I I think I've seen you tweet about it but I don't know anything about it. Honestly, I mean offline we can we can talk about this, but like like for cont this is not about ads, but just for content creation like the being able to do what you're doing inside of Claude projects. Um and and just I have a board set up that's like it's just my template for writing newsletters. This may be off topic so you know.

No, no, go. It's great. Um a template for writing newsletters. So it has context like it's just again it's just context education examples. I've given it five YouTube videos on how to write good email newsletters that are from other people. I have given it 10 examples of my newsletters and then I have given it a voice note of me just like brain dumping what I want the newsletter to be about and like it and then I just pull it into the Poppy the like Claude chat through Poppy and I say take the education on how to write good news newsletters take the tone of how I write newsletters and write a newsletter about this 3minut voice note and it's scary how like good it gets to a point like obviously it's not going to oneshot it but like take gets me to a point from 2 hours to write a newsletter to 20 minutes to write a newsletter and that's like every time I want to write something new I just voice not it or I just you know you can have the the system set up and that great for content creation also great for I'm quite certain I'm quite certain that I haven't tapped this I'll tell you what I'm trying to do next you tell me if Poppy is the right tool and I think this is off the topic of ad creative but there's plenty of people who are creators here I think either way it's a good way to think about this problem I I realized recently when I released a podcast episode like this one, right?

Um, I promote it in multiple places and then I have little bits of copy I need to put with it. So, for example, I need to have a title for this um for YouTube. I need some thumbnail copy for that. Um, probably for three different thumbnails because uh you can test you can split test your thumbnails now on YouTube. Okay. I want tweets and LinkedIn posts promoting this episode. Um, I know which tweets of mine have done the best uh in the past, right?

So, I can feed that context. I want um blurbs to put in the bottom of my newsletter um as well. So, this is put something I put in the bottom of my newsletter is recent podcast episodes um and blurbs for that. Okay. So, there's at least five different unique pieces of written content that I want connected to this. What I want to be able to do and what I'm working on building out now is to be able to download the transcript.

Now, I don't even have to do that with Poppy. It sounds like I literally just give it the video um and um and turn that turn that transcript of that video into at least version one of all of those pieces so that I don't just have to do that anymore. I can have somebody else actually build it all for me. Does that sound right, dude? Dude, I'm telling I mean people can listen to this and think that I'm like on commission for Poppy or something.

Uh I'm not sponsored by them. Uh it's it's just a killer tool and like that is the perfect use case. Um it was it was actually built for content creators, not um not advertisers. Uh it was built for content creators to like help them repurpose content across different platforms. So like my YouTube titles come from there, IDF thumbnails come from there. I have a board inside of uh Poppy that basically takes my tweets and like I've given it a load of examples of you know how to speak in like Twitter language is a lot more informal and a little you know less baity uh and turn it into like LinkedIn language is a little bit more formal.

You need to put a little bit of more I mean at least I put more thought thought into the hook and like explain things a little bit more detailed and write more formally. Um, and it can take tweets and turn them into LinkedIn posts. Uh, because I've just given it a load of examples of tweets, LinkedIn posts, and how I've repurposed them manually. Again, just giving it good examples and why, bad examples and why. And like you can set up a board like that for each like repurpose like repurposed piece of content you want, whether it's Twitter to LinkedIn or like YouTube to newsletter or, you know, YouTube to whatever it is, Tik Tok script.

Um, it's it's really good uh for adding context. I mean, it's stuff that you can do in Claude. It just makes it a lot easier when you can turn it into a board that's and it's that's visual and uh is like set it up once and now every time I want to do a newsletter, I can just come in and voice note and my my newsletter's pretty much there. I mean, it's a good example of the whole thing because what it's actually going to do is make all of the content better too, right?

So, like in the sense of like by giving it better context, better examples, it's going to take your idea. Now, you have to have the idea, right? So, of course, people are going to go out there and they're just going to AI slop repetition of ideas that other people have, right? They're going to take your content and turn it into their own ideas. They're going to take my content and do that. They're going to do that with all kinds of people and they're going to build big followings replicating other people's ideas.

Fine. Um, so it goes. But I think, you know, probably as that happens, hopefully more original thinkers will rise up to the top. I don't know, maybe not. Maybe it will just be whoever is best using the tools to do that and that's just the way the world goes. Fine. Fine enough. Good. Good for them if they make a lot of money doing that. In any case, um theoretically you need to have somewhere underneath that a good idea and then you can take that idea and syndicate it and and um and express it more clearly, better in the language of the platform where it makes sense and more effectively across all those things.

Like that feels like a real win. And again, it takes that same thing which is like how do you just use these tools as well as possible. It's very easy to see how that gets made into a creative process as well for your ad creative etc. which is like at the end of the day, you still want a real person doing something really authentic and behaving in a real way and getting a real reaction to communicate that message most effectively on that channel, but you might be able to get a really really long ways by helping with um research and prompting and and script writing and things like that in the process.

Right? So, it's the same idea as creating a piece of content for you and your newsletter and me and whatever else as opposed um as it is for for an ad creator. It's just about knowing who the audience is, knowing what the channel what what kind of content performs best on the channel and then um accelerating all of that output based on all of those inputs. Yeah, absolutely. And just to bring it back to ad creation, what we were saying earlier, like you can literally if you don't have any documentation or you don't have any context and you don't have like a load of tweets or LinkedIn posts or whatever it is uh or SOPs for um a model to reference like we were saying earlier like you can just ask chatbt say I am struggling I I need some context for my LLMs but I don't have any process documented.

I just know it in my head. give me a list of questions so interview myself so that I can build a context library just go and record that podcast make it an unlisted video on YouTube and then you can turn like you can make a document on research on that you make a document on ideation you can make a doc document on editing and just use them uh as context for the prompts that you build for clients or like for strategy um because then you have the you know the context of the brand and then also the context of how to do research or how to write scripts or whatever be from your own process.

Uh as well as just pulling in context from other people like you know go and pull in podcasts like this. Go and pull in a long form newsletter. There's nothing wrong with taking other people's um No, that's the whole point of that information. Exactly. So people uh are able to make better ads um or whatever the whatever the the goal be. So yeah, I mean there really is no excuse. Like I I I do think there's going to be value.

Well, there already is value in in a handful of the tools. I think the tool is going to get a lot better over the next 3 to 6 months. In the meantime, I really think there's a there's a a good 8020 to be found in building a solid prompting and context library. Alex, I've followed your content for a while. Um, and have have seen what you've been saying for a while and have noticed how thoughtful and interesting and wellstated your content is for a long time.

Um, and then in the leadup to this, thanks to a couple of things, I started paying closer attention to it than I had before. Right? It was coming across my feed. seeing you in these places but all that uh it's great your content is great I don't know how much of it is being generated with the help of AI these days or not but u but it's great and uh I really would recommend to anybody watching or listening go sign up for Alex's newsletter um which is going to pull in a lot of this kind of information make it really tactical for you really um really uh usable for you go follow Alex on X um are you on LinkedIn as well are you active there I am on LinkedIn I'm more active on X and I do want to make the point that I write every tweet out on X by hand Uh there is on my right hand I write it I type it out.

Uh so if you want to get it straight from the source uh that's where to follow me. There is no AI involved there. Not yet. Great. The the links to do all that are in the show notes for this episode and I really want to encourage you to do that. I found you to be one of the most helpful voices on this question. Like I said this sort of mix of the super optimist, super AI forward person not scared really leveraging it and at the same time um I think not overselling it.

I think it's been a really hard balance to strike and uh in my view, you've been one of the best at doing it and so I really appreciate you taking the time. Share that with others and again for anybody watching or listening to this. I'm still there. Go follow Alex's content. It's really really good stuff. Thank you, sir. I appreciate it. Yeah, man. Great to talk to you. Uh hopefully we'll do it again soon. I mean, that was a great conversation, wasn't it?

Alex is so good. You've got to follow up with him. Go to his X profile. The link's in the show notes. Sign up for everything he does. Follow him on on X. Like, go get his newsletter. just an insanely helpful guy, insanely helpful voice in this conversation. I've just been really, really, really helped by him. Uh uh so big thanks to Alex for that. Subscribe to the show wherever you're watching, listening. If you like this episode, you're going to like a lot of the episodes that I do.

It hits a lot of the big stuff. I've got Brett Curry coming soon to talk about YouTube ads a little bit more. That's a really interesting space right now. Brett spent a lot of money on YouTube. He's going to be really useful to you. I've got um Shireen Alber coming soon, Nate Legos coming soon. All kinds of great stuff in my profit monsters series. Um, you're just not going to want to miss it. If you're trying to build profitable e-commerce businesses with ads, man, you're going to like a lot of this content.

Subscribe right now. Don't forget to follow up with my sponsors. That's Behind the Scenes Studio to be an extension of your team for graphic design and video editing for your ad creative. No matter if you're developing with AI or not, they're going to help you push forward with AI forward ways. btsstudio.co to get them going. Um, and uh, and reach out, make a call, just see if they're right for you. Just schedule a call with them. and Intelliggeems, which is the worldleading CRO software because it's actually focused on profit.

Go to intelliggeems.io. Um, and use the code ferris 20 f a r i s20 to get 20% off your first three months today with intelliggeems. Incredible tool that I love, use all the time with my clients. Uh, so uh you can see everything I'm doing at afgrowth.com, including if you want to try and work with me, you can reach out there. I think right now I don't have any space, but you should still reach out to me anyway. Get on my wait list um and see if maybe there's some conversation or if I can send you to the right person.

Email me at podcastfgrowth.com. I'd love to hear from you. If you have any questions, thoughts, something you want me to hit on a future episode, please email me podcastfgrowth.com. You can also follow me on xanderjferris. I'd love to hear from you. Any public conversation around anything I say is always something that is great, including if you're disagreeing with me. That's fine. I can handle it. Argue with me. Tell me why I'm wrong.

Thanks so much again for watching, for listening. I'll see you next time. [Music]