AI Slop Or Original Thinking?

Well, welcome to the AI Marketing Companion. My name is Sandy Carter, and I am so excited to have you here today. I have a phenomenal guest, Simon Davis. He has turned a game studio into one of what I think is the most interesting AI companies in marketing. He is the co-founder and CEO of Where Mighty, the team behind Secret Sauce, a platform that learns your brand once and keeps every asset on brand at scale. Now, before this, what I think is really cool because I used to be a gamer, Simon shipped more than 45 games to millions of players with partners like Disney, Razer, Apple, and his team hit a brief that needed 25 million, 25 million on-brand assets, discovered that no tool could hold a brand that long and built their own. So what we're going to do today is we're going to talk about what he's doing around brand drift, content that you can actually— that nobody can actually use, and what happens to brand integrity when agents start publishing on their own, which we know is already happening. Simon, welcome to The Marketing Companion.

Hey, Sandy Carter, thanks for having me on. It's very exciting to be here with you.

Now, before we get started, I do want to do a call out to our sponsors, and we have two of them. One is SEMrush, uh, and again, my company uses both of these products. Uh, one is SEMrush. We use SEMrush to make sure that we are optimizing everything that we do today for AI as well as for SEO. And the second one is Brevo, because I actually just sent out a newsletter this morning, um, and it is so easy, so fast, really has a way to get to your audience like no other has. So if you haven't tried SEMrush or Breville, make sure that you do that. Okay, so Simon, let's get going. So, um, I think it's interesting. I don't know if you knew this or not, but I used to host a female gaming group. We had over a million female gamers. So this was probably back— it was actually back before I had kids. Um, and I still love gaming. My daughters love gaming too. And you spent years shifting, shipping games to multiple people, millions of people. What was the moment that you realized that the real product was the AI production system under the game and that the studio needed to become Whereamighty?

That's a, it's a very fun question. So we had been building this AI tooling for 3 years. We started in 2022. We started playing, I think, with Stable Diffusion the week it came out because, I mean, you mentioned 25 million assets. We were trying to really hit that scale. And we were really struck. We realized we couldn't do it with humans. It's just we were not a big team. And even the biggest team in the world couldn't do that. And so we started playing with early AI tools. And we found that they tend to drift very quickly. And so we invested time and effort in figuring out that problem. And we fast forward 3 years, and we had something we're very proud of. This is 2025, and we did a demo to someone who works, won't name them, but one of the biggest advertising agencies in the world, one of the biggest groups.

Oh, wow.

Yeah. Yeah, he's a friend of mine, and he was like, wow, this is amazing, can I have access? And I was like, no, this is an internal tool. Like, we've no way of being able to support you and run this. And, you know, I ruminated and I thought on that more and more. And then we kept getting more interest from people we were just showing it to. And we realized actually that this was a much bigger opportunity. This is a tool that could help not just our business, but literally any business out there that has any sort of assets, be it branded assets, social media assets, any kind of content needs, we could technically serve them. And I realized that was actually a much, much bigger business. Speaking personally, I've been making games for over 20 years. I've had a lot of fun making games, had some success as well, but it felt like maybe it was the right time to try something new too. And plenty of gaming companies have done very well in B2B. Slack was a B2B company, sorry, was a gaming company. Discord started life as a gaming company as well. Flickr was another one. App Loving, Unity.

I didn't know that Slack was a gaming company. I knew Discord was.

Yeah, actually the founder of Slack, Stewart Butterfield, also founded Flickr, uh, and it had like a— he had an MMO that I actually played, and the MMO didn't do that well. I really loved it, but the chat system from that game and their internal chat tooling ended up becoming Slack. So it's a rich history that we've tapped into with this.

Really fascinating. Well, I'm a big gamer, I love it, and my— both my daughters have inherited that love of gaming too. So I think it's kind of cool that you really started you know, in that gaming world, because the scale and everything too is so powerful. Now, when we were just talking, you kind of mentioned that 25 million problem. Your origin story is really that you got a brief that needed 25 million on-brand 3D assets, and no existing tool that you could find could hold a brand that long. So Walk us through kind of what happened there. What specifically broke when you tried to scale it and that consistency as well?

Yeah, so again, like we're going back to 2022. So just the very first gen AI image products really coming out. So it's the first version of Stable Diffusion. Midjourney is around the same time. And so you can start to see the promise of what's happening. But up until that point, everything had been manual. And so, I had the privilege of working at some of the biggest gaming companies in the world, and they would have literally thousands of people in developing countries hand doing individual items. You would use AI for limited purposes like generating trees and rocks, but people don't really look at them very closely, so they can be a bit broken, a bit janky, and whatever. But the idea was that every player hero should be unique. And based on all the different variables and combinations, that was potentially 25 million 3D models. So we were forced to start using AI tools because we were like, this is the only way possible. And what we found, to your question, is that you might get one asset you're happy with, and then when you regenerate, like, something just breaks. So like, I don't know, the clothing is all of a sudden wrong, or it's got— you know, the early days you had issues with fingers, right, and teeth looking weird, and they're like 9 fingers or whatever, or like the shoulder would be off. Yeah, and even if you like regenerate the same asset, all of a sudden they start to change, like the faces start to drift and get artifacts. And so it just didn't work. But we knew again, like, this was literally the only way that we'd be able to achieve this. So we really set ourselves to solving this problem. Um, but it was born out of necessity, you know. We didn't, we didn't set out to build a B2B company or to tackle this problem, but it was a genuine problem we had. And by the time we realized everyone else had it. And so it's very exciting to be at the forefront of this.

So if you're a marketer and you're listening to this podcast right now, um, what of this is applicable to like a B2B marketer or a B2C marketer? Um, what in the story makes it really interesting that they should really pay attention?

Yeah, I think the Part they should pay attention to is that we cracked the brand problem. So if you have, I don't know, I'm just picking a name here, right? But McDonald's is doing a new campaign and McDonald's is hypothetically using Secret Sauce in this scenario. They could instantly spin up ideas for campaigns that were on brand. So you could say, hey, we're doing a summer campaign and like the agent has memory of what the McDonald's brand is, what the tone of voice is, who the consumers are, what the values are of the company, what specific objectives the company is trying to hit, what dos and don'ts, maybe preset layouts. So it could be that, hey, our creative always has the placement of the logo here, and down here we have the copy, and in the middle maybe we have a family scene. And so it would start to generate variants of that. They're on brand, they're actually ready to ship instantly without having to rework, and they're basically ready to be approved by you. And so You know, AI has created a bit of an ROI trap. And so that's essentially what we're focused on solving. So you have the promise of AI and we're helping you kind of live that reality. There's a lot I could go into on the subject. ROI trap is something I'm very passionate about.

Well, let's talk about that then. So if I— so I know a lot of marketeers who listen on here don't work for the McDonald's of the world.

Yeah.

And you said you made You, you, this is what you said. You said AI-made content generation can be fast and cheap, so now teams can produce an infinite amount of it. Um, is that what your product, is that what your product helps with? And how does the output often make a marketing team slower? And what's that hidden cost look like inside of a real org? So talk to me a little bit about, you know, who your product is for. Is it for the big guys or is it for, you you know, everybody, and what, how do you help them get out of this ROI trap that exists out there today?

Yeah, so our product is used by companies from one-person food companies in Asia through to fintechs in the US. So it's kind of really spanning the gamut of both size and complexity of business. But, you know, there's something I call the triple bind, which is what everyone is facing with AI now, and I'll break this down. So the first part is that you have increased expectations. So in the era of AI, every business leader is under pressure to generate more content than ever and to cover more surfaces than ever. And so I had a meeting literally with someone today in London who was like, hey, can you help us create content for Chinese social networks? Because now this is a focus for us. And in the era of AI, everyone is trying to cover more surfaces. The expectation is that you should be able to do this. So then the second part of this bind is that, well, there's only two ways you can do that, right? You can scale with AI, but off-brand AI will create stuff that doesn't look like your brand, and so diminishes your brand value. Or you can throw money at the problem, and you can have thousands or whatever, tens or hundreds of humans just churning out as many assets as you can. But in both scenarios there, you are increasing either your spend or you're damaging your brand. And so actually over time, your revenues, your profitability go down because your input costs are going up or you're devaluing your brand. And then the third part of the bind I call is almost like a, it's almost like the brand equivalent of the Manhattan Project where every brand owner is racing to be the first one to actually accurately deploy AI and be able to scale, you know, hundreds or thousands of creators every month. And the one that does that within their vertical is poised to then be able to validate hundreds of thousands of different variants and then find a creative that really works for them and scale and capture the market. And so that's where I think we're at today. And I think the ROI trap, which we're talking about, is that, yeah, you can create loads of slot, but if you don't devalue your brand, then you end up spending loads of time or money. Often people don't differentiate the two, trying to rectify the AI output. Another example of this is like a business that— I'm going to use a hypothetical. It's somewhat hypothetical because, you know, we know businesses that do this, that spend loads of money on creators, right? But only say 2 or 3, let's say 2, 2 videos out of 10 are actually usable. Like if you're spending $100 per video, then actually your effective cost per video is $500 plus like the review. But people are not really thinking of AI in these terms, right? They're just thinking of like the input cost. But the real human cost and business cost is much, much higher.

So you know what's really, I think, interesting about that? I don't know if— did you just see an ad that Nike put out? People were talking about it because it used some of the AI verbiage, you know, like, not this, but that.

I heard about it. Yeah.

And people were all over it. And then there was another ad where I saw people were talking about trust, and the company was talking about trust, and they were using AI actors, and one of them had 6 fingers. And so the comments came in like, how can I trust this company who can't even use a human? They have someone with 6 fingers, some alien. And so there is a lot of risk to doing this, you know, to having, um, having a— having damage to your brand. And it doesn't really matter how big you are, right? Whether you're Nike or whether you're the smaller company that used the person with 6 fingers. So you are, right? Do you agree with that?

I do, 100%. Which is why I talk about this a lot, like generic AI. So, you know, I'm just, I'm not picking on anyone, but like, I know ChatGPT Image or Gemini, these are like generic big models. They are fine for creating one-off work, but they will not look like your brand. They will not retain what your brand's memory is. And so, like, talking about Secret Sauce, we have something called Brand Guard. So there's computer vision both on the input to check it's on brand, and once you've created your asset, on the output to make sure it's on brand as well. And so, like, they don't have guardrails built in either. And so, yeah, it doesn't surprise me if you're using a generic tool that's there to create one-off assets. I think that's fine for, like, maybe ideation or student experiments, whatever. But like, as a legitimate business, like, I wouldn't put my personal brand or our company brand in the hands of like a tool that's built for everyone. I would use, you know, a solution that's either been built with my business in mind or built with our specific use case in mind. So yeah, no, I'm 100% with you on this. Like, I think it's a very bad idea.

And you know, it is interesting. You talked a lot about drift too, you know, where the AI output slowly wanders away through your brand, through small inconsistencies. And I don't know if you, um, saw this weekend, um, Dwayne Johnson, you know, the actor. Yeah, they took a picture. Yeah, The Rock, that's right. They took a picture of him and then they just repeatedly asked AI, just draw the same picture, draw the same picture, draw the same picture, and it showed you all the pictures Even when you said draw the same picture, give me the same image, give me the same image. Did you see that where he ended up like—

He turns into an egg by the end of it.

Yeah, he looks like nothing. He, in fact, I think one of the last ones he was wearing some sort of blue like Marvel costume, which was totally off where the first picture came from. So these inconsistencies matter to your brand for sure. So, you know, if you think about that, how does your tool, how does it make sure that you don't slowly get away from who you are and what your brand stands for, right? That's so important.

I mean, very simply put, we have what we call a brand brain and it's a system of record for your brand. And so it has the logo, it has subsections for any products you might have. So you might have, you know, I've got my Secret Sauce Hot Sauce here. You may have your hot sauce and it's got your brand captured for that. Um, and it has strict rules governing how each one is applied. And so when it keeps regenerating content for you, it's not doing what the, what the example you described is where you're using the previous photo and then generating a new asset based off of that. It's always referring to the source of truth. And again, like it will check the input from the source of truth when it's finished working, it will check the output. And if the output is wrong without involving you, the agent will then go back and redo their work and keep redoing it until it passes the BrandGuard check and then say, okay, like, I think this is ready for you to review. And look, it's a machine, it's automation. There are still times where it comes out and you're like, no, I don't like this. This is for whatever reason, like, it's not on brand, or visually I don't find appealing, and you'll send it back again. But, um, that's how we do it essentially. We have a system of record which is maintained and you can always update it. You can say, okay, this season my brand is not red, it's blue, or we launch a new product, we've changed our verbiage, and it will adapt to that too. But it's all about that brand brain.

Yeah, that brand brain. And people are focused a lot now on the brain, right? On like, how do you create this AI brain that can help you drive everything? But, you know, I just did a session for a group of marketeers directors of marketing, campaign managers, and some CMOs. And it was fascinating to me how many of them— to those, Simon, you know, you talked about you want to make it consistent with your brand book, your brand guidelines, that sort of thing. It's interesting how many of them really had not completed a consistent version of what they want their brand to be. What their real value prop is. And I think part of that is, you know, AI is causing them to change up a lot, which you really shouldn't do with your brand. But I'm just curious, what kind of, you know, suggestions do you give on that? You talk about brand as a system, right? So Secret Sauce goes out and it learns the brand from a website, from a brand doc, holds it across. But in order to do that, you have to teach it what your brand is. So I'd like to first start out with, how do you advise, like, if a brand comes to you and doesn't have their act together, doesn't have that brand doc and that mood board, all those things together, how do you advise them to start? Or do you say, hey, go do that and then come back to me, um, and I will help teach the machine the brand and what part of, you know, and then you can also talk to us about what part of the brand turns out to be the hardest. Um, to get right?

Yeah, so it's a good question. If you don't have a brand, it's not ideal, but the agent or Secret Sauce will speak to you and just start asking you clarifying questions because we have, you know, an outline brand bible just like as a template in case you don't, and it will start to populate it. And I always tell people, like, you want to be spending 80-90% of your time feeding it information and only 20% or 10% of the time actually generating content. Because the more time and effort you put into feeding, the faster and the more accurate the content will be and the less money you'll waste on like failed generations or stuff.

Okay, let's pause there 'cause I think that is really important because, you know, a lot of people, they even do this with Claude and ChatGPT, right?

Yes. Yeah. I know.

They quickly put in like, you know, give me something and then they get it out and it's like garbage. And so your point is that for your system, as a brand leader or marketeer, you should spend 80 to 90% of your time feeding the system, populating with your brand book and your guidelines and what you want that feeling to be, and only, what, 10 to 20 on the actual generation of content?

But to clarify, like, the initial setup is maybe 20 minutes. But even when I'm generating, say, a video— so I built a video today just as a sample for a potential client. The amount of time I spent actually, like, generating the video was probably, like, 20% of the time. And the other 80% was, like, me saying, hey, here's a picture of this— because it was an outdoor thing— there's a picture of this place, these are the type of people that go to it, and basically just giving it as much context and information as possible. And then saying, hey, don't generate video, storyboard this for me. And then it'll go and generate 6 assets. And I'll be like, OK, change this in the storyboard, change that. But actually really going through that process of planning it— I always say this, and it's become a bit of a cliché. And other people say it too. But AI is like the world's most powerful intern. But at the end of the day, it's still an intern. So the more information and context you feed it, and you really make the instructions super clear, and you give it a very clear idea of what you're going for, they get amazing results. But if it's like, hey, give me— I'm just picking an example— give me a 30-second video for this theme park, and it doesn't know anything else other than like, here's the website for this theme park, you're probably not going to be happy with the results. Like, maybe, you know, I call it the prompt lottery. Like, maybe 1 in 100 times you get something great first time. But, um, I don't think luck is really a great business strategy when you're trying to allocate your time.

Well, I think this is a really important point because I think that if I look at all the data out there, BCG, McKinsey, et cetera, they talk about 20% of a company's success, maybe 30, depends on the technology. The rest depends on people and processes. And essentially, that's what you're saying. You're saying our technology is going to kick out the content, but it's going to be only as good as that 70% to 80% that you give it inside, that your people give it. So I know there's a lot of people out there who are really worried about their jobs and their roles and what do they do with AI. I think this is the reason that people still matter. You know, I wrote the book called AI First, Human Always, because what you just said, Simon, is the people content, how the brand wants people to feel. You know, you just talked about the setting, all of that. Is coming from the human. The AI is just taking that, using that as the context. So I do feel like people have such an important part of the success, especially in marketing and branding even today.

No, 100%. I mean, when we look at the product today, you know, I can sometimes take a look at like kind of who's having success with it and they'll be in touch. And by and large, anyone that has a strategy background or is a strategic thinker, absolutely crushes it on there.

Oh, interesting.

And that's what I love about it. So, like, if you can, you know, like, there's so much, I call it design administration that goes on. So, to your stat, I was reading somewhere that something like over 85% of agency budgets or production budgets are spent on stuff like resizing, like doing localization. I call that design administration. And so that's the stuff we really want to, like, automate as much as possible so that people can spend 80% of their time thinking of an amazing campaign or a new way to engage their customers or new strategies for reaching their audience and, like, really iterating on that and not, like, doing the boring stuff of, oh, we're going to make this, I don't know, 15% bigger or invert this or whatever. Like, that stuff be handled.

Yeah, I think that's really important. So I think your important takeaway is that you have this amazing— you have an amazing tool that can help you do this. But again, it's helping you do it. It's not necessarily doing it for you. You still need to be the thinker, the strategist, if you would, as you're moving forward. What is the part of the brand that turns out to be the hardest to really capture with an AI tool?

Well, you know, AI has not gotten to the point where it's a mind reader yet. It's still some way away from that. So if people don't give it explicit instructions or like clear, like kind of guidelines, the tone can be a bit wrong. So I'm very particular about language. I read a lot and I've written a decent amount. And so I'm not 100% satisfied with our copywriting yet within the bot, and I think that's still hard to get right. I mean, everyone who uses AI regularly has encountered asking it not to use em dashes and having it come out with em dashes. That's really the tip of the iceberg. And you mentioned it's not XSY, like these kind of AI tells. We do a pretty good job of getting most of that out, but copywriting is not where Secret Sauce excels. Secret Sauce excels for visual development and visual creative and videos and anything like that. But I think copywriters are gonna be safe for a while, yeah. Great copywriters are worth their weight in gold. And actually they, again, like the strategists, they tend to have very good results because they can communicate very clearly and communicate that artistic vision. I think for me that's still the hardest thing to really get right.

Yeah, I think that's probably true. And you know what's so interesting is I was reading, I always go on LinkedIn, you know, just to get caught up on the news. And today I had like 5 different articles I was reading and the statement was, and this stopped me dead in my tracks, or this stopped me cold. And I was like, oh my gosh, it can't be that all these people are saying this at the same time. This has to be coming from, AI and things like that, you know, really, really, really annoy me. I use AI to kind of do a check. I'm not the best on grammar and I am horrible on spelling. So I use it to do a check on that sort of thing, but I don't allow AI right now to write for me, like write my articles unless it's that, you know, that misspelled word or something like that. Now I know that marketing is moving more to agents. I've actually just been in conversations with several companies who want an agent to generate, approve, and publish on their own, which scares me because then who owns the brand integrity? Because now you don't have a human in the loop, on the loop. They're nowhere there. So what guardrails do you think will separate teams that trust their agents from the ones that just delegate You know, you go out, generate, approve, and publish, you know, have at it. What do you think about that?

Yeah, it's funny. You know, I go back to my previous point about it's an intern, and I use a lot of AI, every tool you can imagine. I wouldn't trust an agent to publish today. Maybe we'll get there, but I actually don't think that will happen anytime soon. And also, like, agents are not gonna be aware of the zeitgeist or things that are happening around it. And so it could be that, just use an example, right? But one thing that's fine to publish one day because of an event in the news becomes like a very bad idea. Agent's not going to catch that. And so like, I would— anything that's going live with your brand attached, I'd always make sure that this has to be a human that reviews it at the end. We have, um, we're still arguing over the name actually. We have agents that I call Relentless Agents that work for you 24/7, and they scrape the news, and when you log in, they'll come in and be like, hey, like, we've got these 10 ideas for you to publish today, and they'll go off and just do the creative without even being involved. At the end of the day, the idea is that you receive that, you vet it, you give it feedback, and then you decide if you want to publish. Um, yeah, I, I, um, these agents can be very smart, they can be very capable, they can do a lot, but I wouldn't trust them with, uh, you know, damaging your brand can be a multimillion-dollar outcome, right? I wouldn't trust them with just publishing stuff. Yeah, I had— I'm going to share an anecdote I've never shared anywhere before, but like, I had an agent that was running on my machine and I just had it doing some analysis on, on some, some social media posts. I was like, well, just look at this, like, give me feedback on what's working, what isn't working. And without me approving, it just went off and posted something that was completely false. And I was— yeah, And I think that just shows you, and that was not long ago, it was a few weeks back, it shows you how primitive these agents are and how bad they are at sticking to guardrails. And so I would not under any circumstances recommend anyone trust an agent to approve and publish anything.

Yeah, you know, it's interesting because I have had friends who I really admire. And in fact, one the other day, I don't know if you know, right now on social, there's a picture of this building going around. It's supposed to be like the newest, tallest building, but it's fake. It's a complete AI-generated building. And I've had several friends that I really trust who've posted that picture, and I've had to go to them and say, you know that this is an AI-generated picture. And then a couple of people, I don't know if you remember when they had the rocket shoot off and they were showing pictures from, this is a picture from an airplane, and it wasn't accurate. People were like, ah. So yeah, you have to be really careful today for sure. So to close this out, Simon, I'm really interested. So, you know, we've got all these amazing marketers who listen to the podcast. And, you know, if you're sitting there and you're feeling behind, what would be the one move that you'd tell them to make this quarter to focus on getting their content working for them and even using AI to do that? Instead of drowning in like a whole sea of assets that they're gonna have to just throw away? What would be your advice to them?

Well, I'd say just start. Like, you can just do things. AI seems very scary, but, you know, whether it's Claude Codex that can now like do computer control for you and do like simple tasks like updating spreadsheets, or sorry, not Claude Codex, OpenAI Codex. I was gonna ask you if it's a Claude head purchase. Yeah, Claude Codex, sorry.

Yeah, yeah, yeah, yeah. But I love Claude. I use Claude all the time.

Yeah, well, Claude Cowork, which is really good. But like, just start with one and start with something very small. Like someone the other day was like, oh, you know, I wanna make an app, but I don't know what. I'm like, well, just speak to the AI. Just tell the AI, hey, I wanna make an app. Like, give me some ideas. And just through natural language, right? And just these baby steps, they start to compound. And before you know it, you're doing all kinds of things that you thought were unbelievable, that felt like magic. I will also caution against the trap, which is that when you really start to unlock these tools, and you can be so, so, so productive, there's that temptation to just keep going and going and going and doing more. Before you know it, it's like 3:00 AM. And so, I've never worked this many hours. That is the paradox, right? You can do more work than ever, and I'm probably like, I don't know, 5, 10 times more productive than I was a few years back. In terms of output and being able to cover different topics. But I'm also working longer hours because it can be quite addictive. It's not like it's not fun as well. You feel very powerful.

I agree with you. I think it's such an exciting time to be alive right now, be able to experience this, be able to use tools like your tools, try them out, see what the potential is. And you know, there's no penalty. You just have to put some time into it. Simon, before I let you go, could you, also let people know, like, if they wanted to try out their tool, can they go try out your tool? Um, can you kind of give them some guidance on, on where they would start?

Yeah, so, uh, I would tell everyone— you mentioned try— to go to trysecretsource.ai. So you have to put the word try in the front if you want to give it a go. So trysecretsource.ai and just play with the platform. And I answer every email that I get, so if anyone's got any questions or they get stuck Wow. You can contact me, Simon at— sorry, simon.davis@trysecretsauce.ai, and I'll answer any queries on this as well. And yeah, like, just have fun, play with it. It's free to start and see what you can create.

Super. Well, thank you. Thank you, Simon. Thanks for joining us today on the AI Marketing Companion. Thank you all for listening to us. If you do have questions, I send out a newsletter. We've got ways that you can ask us questions as well. And just remember our two sponsors who brought this to you today, SEMrush and Brevo. Brevo for doing your newsletters and your customer outreach, and then SEMrush for helping you with that AEO and SEO that's out there today. Thank you so much and come back and see us again. Thank you, Simon.

Thank you so much, everyone.