Stop Trying to Fool People: How Movers+Shakers Uses AI to Amplify Creativity, Not Replace It

Artificial intelligence isn't the future of marketing, it's happening right now. AI is transforming the way brands connect, create, and grow. But what does it really mean for marketers today? That's what we explore in AI, machine-made marketing, a podcast from advertising week. In each episode we dive into the tools, the strategies, and human stories behind AI-powered marketing. Join us as we explore the power of intelligence that's not just artificial, but transformational. Everyone wants viral AI content, but most brands are doing it wrong. On this episode, Jeff McDonald, director of Innovation and Technology at Movers and Shakers, joins us to unpack what separates AI gimmicks from true creative acceleration. Jeff sits at the intersection of social first storytelling and AI tool building, helping one of the internet's most viral agencies design systems that compress time without compromising taste. His take? AI shouldn't be used to trick audiences or replace human instinct. It should function like special effects, enhancing the story, not pretending to be the story. We hope you enjoy this episode. Jeff, welcome to the show. Thanks so much for taking part today. Thanks for having me. So for the uninitiated, let us know what do you do at Movers and Shakers? Absolutely. So, Movers and Shakers is a social first agency, and we really concentrate on helping clients go viral and creating content for them for the social web, specifically video. And I have a very interesting role where I am kind of at the intersection of both social, but also like AI innovation. So my job day to day is to help enable the team through innovation to do their job better, whether that be creating them new tools or finding them new tools. And luckily for me, my job has mostly been focused on creating new tools lately, which is really exciting. Excellent. Well, let's start with, and I hate to do this, but let's start with what everybody does wrong. So from your vantage point, you know, within innovation and technology, where are brands still misusing AI in their content creation process? I think the biggest thing that brands are doing wrong right now is they are assuming that they can trick consumers into looking at AI content and assuming it's real. Kind of like saying, oh, is this AI? I don't know. Like that's that uncanny value effect that everyone's trying to to overcome and say, like, oh, well, VO 3.1 is this much more realistic? That's not the point. Like we shouldn't be using AI to try to fool anyone. That's actually just going to harvest more distrust in the tools that are eventually going to make it so we're more efficient. And that's inevitably what we want, right? If we're a company. So that's really kind of the issue is instead of seeing it as a way we can plus up what we're doing with our creative minds, with our creative capabilities, we're trying to replace it wholly with AI. And movers and shakers, we don't use AI to create holy AI generated content. We use it to plus up what we do really well, which is to create human driven stories and narrative and creative content. So we see AI more as like what you would use special effects for. I love examples that we have where oh, no one would expect us to have this giant alf cosmetics product in the middle of the street in London because like it's just a social post, right? So of course, it's special effects. And in our case, the special effects we're using is AI. We're using Adobe Firefly to render off this product. So I think with those kinds of examples, you get a lot more leeway with the users with audiences, whereas when you try to come out with completely AI generated piece of content, and you don't acknowledge the fact that it's AI, because you don't do anything fanciful, you don't do anything where it's like, oh, well, of course, they had to use special effects to do this. You're just trying to recreate what you could have done if you wanted to, if you capture it. That's what I think brands are doing wrong. So hopefully what we'll start to see as more brands to start to figure out their AI workflow and see it less as a trick or a magic trick really, they'll start to see, oh, well, AI helps us make this thing a little easier. AI helps us plus up this or, you know, even simple things like AI helps us up-res the video footage, so it just looks better. Like that's a great example that every brand should be doing. Nobody should have poor sound or poor video quality. So sort of the difference between rather than having AI create a sequence where Tom Cruise is fighting Brad Pitt, you could just have a sequence where Tom Cruise is fighting like, I don't know, a giant robot, because we know that would have to be created in some way, be it AI, special effects or both. Yeah, there's reasonable expectations that people have for how a brand creates content, and there's things like, well, if you're going to shoot a video entirely underwater, human, you know, breathe oxygen, so clearly something was done to make it so you could create that effect. I think if we have that reasonable expectation and we meet those expectations, users will be much more accepting of us using these types of tools rather than if we're just trying to do a one-to-one replacement of what we could have captured. You've described tools like, you know, well, just AI tools in general, as sort of a partner that never gets tired. Yeah, how do you operationalize that inside an agency and what does AI assisted ideation actually look like on a live brief? So this is something I was just on a call about and we were talking about how do you find those moments where you can kind of collapse time. So when we look at a normal workflow, whether it be like ideation, content development, like mood boarding, and then creating the content, we'll look at those parts, which are like the what we'll call, let's call them human and elute moments where human is really important in that process and just figure out what are the parts that we can collapse. So for example, like mood boarding, mood boarding is a very important human in the loop step of creating a creative concept. Like you want a human to be the curator because they have the taste. They have the, and that's why they're hired because they have the taste of the talents to understand what looks good. But what we can collapse time on is finding the images that go into a mood board. Like a lot of time, if you think about it, goes into Google searching, the art style, the color palette, the specific image, maybe the angle isn't quite what you want to share with the rest of the creative team or the rest of the project team. And so you want that perfect image. We use AI tools to collapse that portion. But then the human element where it's curating the images to make the perfect mood board, we protect that part. We make sure that that creative, that production assistant has the most amount of human in the loop time that they would have had prior to AI. But we're just collapsing time. We're doing time compression in the areas where we know we can find automation and efficiencies. And that's with things like image search, image creation, especially around mood boarding. So we have a whole entire project that our agency, which is around AI mood boarding and AI story boarding to help collapse that part of the process. But then there are other areas where we don't collapse. Let's shift over to brand agents. If you're building an AI system to think like a brand, what inputs does it need? You know, tone, audience psychology, cultural context, performance data. What separates a real brand agent from simply a clever prompt? Data is the short answer. The long answer is really building in segments of that agent's process that allows it one to allow the human to come in and give the taste element. So saying, yes, these are the top performers. These are the bottom performers of that brand. So try to do more of the top performing thing. That's the data portion. We feed in all of the best videos we've done for the brand and the reasons why they think they're best, those are the best videos. But then there's the human and the loop element, which we give pause sections of all of the AI tools we create to allow for the human to come in and say yes to this, no to this, that helps steer it. So that's I think the main huge element is making sure that there's that human part. And then the last part is kind of a new capability of a lot of these AI agents, which is the compounding learning effect or compound engineering as it's being called in some coding circles. So what we build is a capability once you've directed the branding agent to write you the perfect post copy or to help you write out the script for the storyboard you're making for a TikTok, there's an element at the end once you've created the deliverable, I've made my storyboard, I've made my moodboard and it asks you what went right and it's just a simple prompt, what went right? Or it'll say let's wrap up and at that last step, what happens on the backend is it writes its learnings. It says, oh, earlier on, I should have known that for this brand, they always use this color red. So from now on, when I reference this brand, I'm not going to bring any other colors red or I noticed that my copy was tweaked by the human to say this. And this was the reason why or I'm going to ask the user now as we're wrapping up, why did you change this copy? It writes all of those learnings from each initial each each workflow, each run through the paths through to a learnings document that subsequent agents use every time they do that process. So whereas if you think of a Gemini gem or if you think about a chat GPT, GP custom GPT, they're stuck in time, they don't learn from it. They help you, they help you do the same process over and over again, but there's no learning document that says there that helps it understand why you had to correct it four times to get the prompt that you or get the output that you actually wanted. And that's what we've built into our agents is the branding agent specifically. They have a learning element that helps them stay more on-brand as the human is the curator. And does then that agent take in to account, you know, down the road, performance and things like that to further educate itself still? Yes, it takes in the live data as part of the process. So one of the steps of the flow is, you know, I'm going to branding agent, I'm going to help create copy for a post. Input with the post image or video is here are three variations of post copy based on prior learnings. So it goes in it looks at all of the prior learnings it has both from the training document, the learnings of subsequent workflows you've done. And then it goes in it hits an API that we have with our social media management tool and it pulls the latest post data. So it says of the last 24 hours since you lost use this tool, there's been two new posts. Here's what we learned. And then it subsequently continues to add it to that index. So yeah, it constantly is updating and it's constantly learning from it. And I think that's pretty unique. I know a lot of people who are creating like custom gyms or custom GPs that help them with staying on brand and uploading a lot of branding documents. But I love this concept is that each work unit you do should make the subsequent work units easier and you do that by creating AI agents that learn from your corrections. Because as humans, we learn by doing. And so in order to make AI agents more efficient and better, they need to learn by doing as well. In your experience, what does AI amplify best? Speed, iteration, insight, media performance? And where does human instinct still win out or is at least favored? Human, I could start at the end there. Human insight and instinct in pattern recognition. Like all of the things that you would think that are the most human things. I had a really great conversation with a client. And I said, you know, one of the things I can't replace is that I feed up a bunch of information to a strategist or creative and they pick the best. Because they've got the taste, they've got the pattern recognition. The strategist can look at a bunch of desperate ideas and pick the three that are actually the insights. A creative can look at 150 images and pick the five that's going to make the best mood board. And there's that pattern recognition and instinct that you can't replace with AI. So what we use AI a lot of times the do is to provide options. And what we use humans to do is to provide the curation and taste. One of the things I think is really unique is when we think about how we create tools at Movers and Chakers, we're not creating answer engines. Like if you think about chat, CBT or Clawed, when you put in a request, they're giving you what Clawed and chat, CBT, all of these tools, they're giving you what they think is the most likely the answer you wanted. The 95th percentile, like this is what the answer should be. We build more what I would call Pinterest tools. Like if you typed in best pair of genes in the Gemini, it's going to be give you a top three pairs of genes based on reviews. If you go into Pinterest and type in best pairs of genes, there is no answer. And it's just going to be right now, unfortunately a lot of AI generated content. But it's going to give you a lot of responses about the history of denim, how genes are recycled, a video about how somebody distresses their genes at home. It's just going to be a lot of things to react to. So when we build our tools for both our strategists and our creatives, that's how we build them. We feed a lot of interesting ideas in front of them using lots of separate data sources. And then we allow the human that the strategist to create it to tie the two ideas together and to select the favorites and then to drill down further and say, oh, I like these two things. How are they related? How do I click on that and go a couple layers deeper and actually get the thing that's interesting? Or maybe it's the reverse of this thing that I'm actually finding interesting. So that's what we build. We build kind of like these Pinterest like tools that allow our strategists and creatives to fall a little bit down the rabbit hole and to do it faster, to use it as an accelerant, to get to that first idea, to that first insight and increase the speed to insight and the speed to the creative concept. Yeah, I've read before that you've said that we need more Pinterest experiences and fewer Google experiences. And thanks for clarifying exactly what you meant by that. Yeah. I could agree more. Let's go down that a little ways. Can you give an example about how AI would help a brand then design for discovery, curiosity, participation instead of just those answers? Do you have a specific tangible example in mind that you could point to? I mean, I would think to say, if you were a brand that were wanting people to curate with you and create with you, you need to not give them tools that are supposed to just be answers. They should be tools of creativity. I'm going to use an example that we don't have as a client. So that'll be safe. You know, if you were homes.com or or Zillow, instead of just typing in, I want to see houses that have these exact specifications. This number of bedrooms in this zip code near these school districts, make it so that using AI, they could be creative. They could say, let me sketch what I think my perfect home would look like and then provide examples of what those homes might be. Kind of really allow the user to explore and co-create with you rather than giving them answers and saying, nope, this is what the AI told you. This is the, this is, or you know, for Uber Eats. This is the dinner you're having. AI is telling you exactly what you're having for dinner. Instead of that, make it so you're exploring with the AI so that we are amplifying again, human taste and interests rather than surplanting that and replacing it with just an AI has the correct answer. Yeah, it's interesting. You know, when you were talking about that, you know, in the previous quite answer, I was immediately thinking of, you know, this is what Google, not to pick on Google, but like search typically has done wrong for me is rather than, you know, kind of guide me where I feel I want to go. It's showing me 10 things that are angentily related and I just start going down rabbit holes and then eventually get nowhere. Yep. Now, which is kind of a danger to people with, you know, fresh to be attention spans like my own. Same here. Absolutely. Like I need there to be some kind of opportunity for me to steer the conversation, collect myself back. Also to have breadcrumbs of how I got here. I think that's the other aspect of it that people have it really explored is allowing people to kind of reverse their search and say, yeah, you know, I mean, this is not exactly what I was looking for. And that's what's kind of a Pinterest like experience gives you as you can say, well, how do you know, how did I get to this node? I'm three nodes deep on the history of denim. How do I roll it back to where it's like, here's where the brief was. And so I think that's really unique and kind of like a fun experience in designing what the future of human computer interaction is. And that's where I'm playing right now and it's super exciting. I don't know. How are you raising the AI IQ of creatives and strategists and clients without overwhelming them? I would think that would be a pretty tall order right now. It's everyone's at the different level right now. So we have clients who are completely in on AI. They're using cloud code to do their spreadsheets, everything. And then we've got, you know, employees and clients who are just starting out and trying to figure out how to use AI within their ecosystem within their workload. And the role of raising everyone's AIQ, which is our first initial goal when I came on board, is to say that AI is a good tool, but it's not always the best tool. So raising your AIQ is not just about helping people understand how to use AI best, but it's also helping them identify when they shouldn't be using it. Because there's lots of opportunities for you to say, well, I could bring Gemini in to do this. I can use notebook LM to do this. But it's about also identifying those moments where your human instinct is better or the tool you've been using for years, you're just fast-routing because you've been using it for years and you know all the shortcuts and just understanding where you can find those compounding effects. So when we took on the initiative to raise our staff's AIQ and then subsequently our clients, we really started at first listening to our understanding where everyone was. And then from there, we just started exposing people to ways that they can compress little segments of time. How do you just like cut down your research time? A simple one was meeting notes, very early on saying, no one humans taking meeting notes, unless you really just like taking meeting notes, which we found users who do, we found staff who like taking notes, compressing that back into an AI task and then making it so that all meeting notes write to a notebook LM for the project. So now there's a queryable knowledge source for all meeting notes and everyone has access to it. So anyone who missed the meeting or is new to this agency, they can go into that notebook LM, type a question and it pulls up all of the subsequent knowledge from all the meetings that we've had on the client. There was just a simple little task that we did, but all of a sudden, we found a couple of hours savings for every single person. And we're just trying to find those moments in this initial stage and then we're really hitting the speed ramp now with our AI agent tools where we're being able to find those really big wins like being able to help them with storyboarding and things like that. Looking ahead to the remainder of 2026 and beyond, what does an AI native creative agency really look like in your mind? So what capabilities are given and what becomes the real competitive edge? Well, first and foremost, it's connecting all the tools that you pay for as an agency in the wine interface. That's really the first initial stage. Right now, so many agencies pay for different tools, whether they be creative or strategic, and really a lot of what the work is, is connecting those sources. It's connecting what are our top performing tech talks of the year with the creative who's writing the next tech talk. And right now, that's a process to go download the data, interpret the data, present it as a presentation to the creatives and the creatives get that data back. And we could co-ops that using an AI agent who can in real time go and provide learnings to the creative where the creative is doing the work, be it make me like just a Google doc and they can just in Gemini chat say, hey, at creative agent, what are the top three performing tech talks for the quarter? And what are the key learnings I need to know about the TikTok idea I have? We've collapsed easily eight hours of work there. And so what we're doing is we're trying to find those types of moments. I think a truly connected creative agency looks for those. The way is to connect all of the desperate data sources that they have, all the desperate subscriptions they're paying for an application that they have in the one system. And I think that's kind of the stress that we're seeing on the social or excuse me on the software industry right now is everyone's also having the subiphanie, which is I don't I yes, I need all the subscriptions. I need my sales force. I need the but I want all the data to talk to me where I want it. I don't want to have to look at it in your user interface. I want it to be where I want it. And that's what we're building. The first step is just bring all the data into one user platform and have it available wherever the user wants it. So I think a connected creative agency is doing that. They're bringing all of their tools to tools to the forefront. And then the next thing that they're doing is that they're making it so that the human part of that is amplified the most. So for us it's creating these more Pinterest-like experience for other agencies who are a little bit more lenient on what they you know we create video content. So it's a little hard for us to go all in on AI generated content. But you know if you're a static agency if you're mostly building like re sizes of content for like paid media and things like that. I can totally see that they're more leaning in on how they can use AI to help with re sizes, versioning, AB testing, things like that. And they need to be looking at that part of the workflow as well. Well Jeff, I could talk to you about all this stuff all day long here. I'm very very infectiously enthusiastic about it. So if one wanted to find out more about you and movers and shakers and all of that, where would they go? Yeah. So our website is movershakers.co. You can come check us out and check out our work. We've got a bunch of examples of things that we've done using AI. And then for myself, that's to look me up on LinkedIn. Just look up Jeff McDonald at Movers and Shakers. And I post pretty regularly about what I'm building. I try to build out in the open. Excellent. Well, thanks so much for taking part in the show today. Anytime. Thank you.