How to Fix AI Slop: Better Context, Stronger Brands and the Return of Human Judgement with Pacey.ai

0:00Ciaran Rogers: Welcome back to the Digital Marketing Podcast brought to you by TargetInternet.com. My name's Daniel Rowles and in this episode we're looking at some groundbreaking AI software but also really what's essentially a masterclass in avoiding AI slop and getting back to the fundamentals of marketing.

0:23Ciaran Rogers: So in this episode Kieran is speaking to Evan Greally from Pacey AI. Now, Pacey AI is a really interesting AI platform, but as ever, it goes a lot beyond, and there's a lot of lessons in here about how really we should think about marketing, going back to some of the fundamentals, how we can avoid AI slop, how we can encourage, actually give AI taste, which is quite a controversial point in the interview as well. looking at how we've probably got lost a little bit in marketing because the amount of data sources that we've coming in and so on and the limitations of the approaches for AI and content creation at the moment.

1:04Ciaran Rogers: So even if you go beyond the piece of software itself I think there's a huge amount of lessons in here that we can learn from. As ever we have no commercial connection to the software and I think what they're doing is really really interesting but I've taken a lot away from this just to think about how I use AI on a day-by-day as well. Oh, and by the way, it starts with having AI-driven conversation with plants. So over to the interview.

1:29Evan Greally: My name is Kieran Rogers, and today's episode is all about discovering a brand new thing, what's brand new to me, called Pacey AI. And to help us in this task, we have Evan Greeley from Pacey AI. Hello, Evan. Hi, Kieran. How are you? I'm good. I'm good. I'm excited about this. So just explain to our audience, who are you and what qualifies you as an expert in the AI space? Because it's such a new space, right? But actually, in my mind, you are a true expert. You've been in this space for a long time. So just fill everybody in on some of the stuff that you've done in the past.

2:00Daniel Rowles: So yeah, I'm the founder and CTO of Pacey. And before that, I was the head of creative tech and innovation at Droga5 Dublin, part of Accenture Song. And there I basically led a lot of AI prototyping and AI pioneering in the space across a lot of different categories like banking and hospitality and networks and different things like that. And I've got 15 plus years working where creative and technology meet. The work I'm most known for is making AI people actually want to connect with. And one of the best examples of that is I created a project called The Talking Tree, which started off as a mint plant in my office in Droga 5 that then a lot of people heard about and it grew into The Talking Tree project.

2:39Daniel Rowles: So you had a mint plant.

2:40Evan Greally: Yes. We're talking about a herb here. A herb, yeah. Literal mint plant on your desk. What did you do to this poor little fella?

2:46Daniel Rowles: Outside of using it for a mojito, we took a lot of sensor data in, so like soil moisture, humidity. It was put into its own terrarium kind of environment. So we took everything like photosynthesis levels. And we also connected bioelectrical pads to the mint plant so we could track response to its environment as well, not just the external stuff it reacts to, but how it internally is reacting to things. We fed all that into different AI models. One was more of a causal AI model, which is cause and effect. It's the best way to say it. So something happens, it reacts in this way, and we learn what that meant.

3:23Daniel Rowles: and we all then feed all that into a large language model giving the plant a voice oh how does a mint plant sound it started off as a hollywood valley girl uh which was a strange experience she then kind of morphed into um more of a scientific metalhead because she was sitting beside me and we gave her memory and learning and everything so for me it was more about at the time was how far can you push a large language model when you give it the amount of data and when you give it a self or a sense of being like i love it herb augmentation yeah i was not to love about that that's awesome and then we did the talking tree project which was basically a portable version of all the sensors and everything that you could connect to a tree we actually won a red dot award for design just there a few months ago and it was just about people coming in and getting young people to reconnect with nature and if you could have a conversation with nature which was more about have a conversation it's like an art piece really but it was if you could have a conversation with nature could you you know would it make you more prone to put down the phone and actually go outside and look at the world yeah and it was it was very successful and it was a lot of fun to

4:36Evan Greally: it was a weird one but a lot of fun so you you got together with richard carr and you guys have built Pacey, which that's what I really want to get into. But before we do that, in a nutshell, what is Pacey, Evan? Let's just explore that.

4:50Daniel Rowles: Yeah. So Pacey, you know, a lot of marketing platforms today give you one of two things. They either know things or they create things. And there's very few that connect the whole stream from end to end. So Pacey is a system of action to help you create better content. And what I mean by that is that when you come on to Pacey, you start off with your goals it then builds out a strategy we then look at signals every day across your social media channels competitors what your audience profiles are doing what they're searching for as well if you have reviews like trust pilot and all the way to google and bucky.com we read all those reviews for you every day and we get a sense of of your daily world let's call it and then we turn that into a space that creates action where you click something and it creates content for social media, but the content is highly relevant to that day and that time.

5:42Daniel Rowles: We track all the performance of every piece that goes out and it feeds back into the strategy. And that just means that we have this compounding loop that's continuously learning about you and your brand and it's just a better space for content.

5:55Evan Greally: Tell me, what was the impetus behind creating this Pacey tool? Because I really want to explore what do you think is broken within marketing teams right now and how does Pacey fit into that?

6:06Daniel Rowles: Every client we talk to says the same thing. I have data just everywhere and it's nowhere in front of me. It's like I can't read 70 reviews every day and you know understand what they all mean and I can't look at like Reddit and I can't look at all the comments on my Instagram and I can't watch news. So it's yeah it's an impossible problem.

6:27Evan Greally: Everybody was obsessed with big data that was the big sexy like buzzword wasn't it big data and we were like yeah big data people started flowing it into their cvs yeah i've dealt with big data it's like you've dealt with a lot of stuff that was so big and convoluted you couldn't actually get your head around it look and it's like oh do you have a data lake i was like what

6:45Daniel Rowles: what's a data lake but i think a lot of the problem with that is that it's just overwhelming even when we talk to clients like we got one of them to start counting oh because we have to

6:54Evan Greally: connect that into pace we need to connect all your data yeah and they were going counting and

6:59Daniel Rowles: they got to 14 and i was like just stop okay we have enough there to start with anyways and it's all just all over the place it's in different formats and it's not that these teams are disorganized or anything like that it's just that they've had to buy into so many different tools yeah and different types of data to be told you're a good market yeah well i didn't get into marketing to read six different tools and reviews every day and everything like that i got in to make good content to push my brand to grow my brand and that's why i and richard when we look at pacey it's about giving people back the opportunity to get into the actual business problems the ideas the things that grow you as a business and that's hopefully what looks like pacey will achieve why

7:44Evan Greally: is most ai marketing so disappointing that's another thing i want to ask you so if you look at

7:50Daniel Rowles: tools today and even Gartner's own research 49 % of MarTech and marketing tools just remain unopened and that tells me that everyone's kind of sick of the dashboard life that they've been living and all that data means nothing to anyone and when you push it into an AI system and you're not giving it any good context you're also getting out mundane kind of things at the other end because first off you don't really understand the data because there's too much of it um so no one's making sense of it and then the ai model goes okay but what's important in here and you know when you give an ai model too much context it just averages itself so it's just like

8:27Evan Greally: everything becomes very bland it's actually one of the most important questions i think marketers have in their toolbox right because actually we're constantly looking at dashboards and data and it's endlessly fascinating and exciting and sexy and all those things right yeah but the question we all need to be asking is yeah but so what exactly so what the x equals y or so what that this happened or like what do i do with that what's actionable about it that's that's the key thing and that's one of my big frustrations a lot of the ai tools i've seen just complicate things with more and more information it's like i've already got too much

9:00Daniel Rowles: don't give me more and this is like the thing we strive for with pacey is honest silence so one of our brands gets about 500 signals a day they're a lot bigger they have a lot of outlets but they We have 500 things a day that before Pacey, they were told they need to worry about whether this is reviews and everything else going on. And we distilled that down into a very simple, this is what happened last night. And it could be, there might be, you come in one day and Pacey will say, there's nothing for you to worry about in your market today. And that's the honest silence part. Or you come in one day and there could be 10 things because a competitor launched X and another competitor did something and something bad happened in one of your hotels.

9:40Daniel Rowles: and you're getting 10 bad reviews. So it's about surfacing the things at the time when they're relevant, making sure that everything that gets surfaced into Pacey has to have three or more sources saying the same thing. And whether that source is across the same data source for us or multiple different data sources. And the point being is that now you're focusing on these are the things I need to worry about. And the content gets better because it is off the moment. But then the content itself is another issue because if you feed in 10 brands to Chachapiti and you ask it to generate the same poster for 10 different brands, we'll get back the same thing.

10:20Daniel Rowles: The same bland looking thing, be like this kind of weird luxury level thing with this weird serif font. And it just always looks the same. So it's, for us, it's also about pushy taste within AI. And that's one of the biggest problems is how do you give taste to AI? because I truly believe that AI can have taste, but it's not human taste. Human taste is a unique, beautiful thing. And it doesn't matter if you go to an art museum every day versus you like staying home watching, you know, reality television. You have, every human has a version of taste for themselves. Whereas AI models have the same version of taste across for everybody.

10:59Daniel Rowles: And that's the problem. It just keeps averaging out. But taste is about, is the opposite of averaging. It's about giving the system a point of view before it even generates anything. And we do that through the data that it takes in. So this is new data every day. So that way it's like not working on the same data source that it's always been working on. And then we also, when we create a piece of work, we don't feed it into a model with the brief to critique that piece of work. We say, you are a person, this audience member walking down the street and you see this piece of work. or you're scrolling on your phone and you see this piece of work would you stop and then we we don't score that it's because getting llms to score anything is they'll give you a different number every time for the same question it's more about true false yes i would yes or no i wouldn't and then the nuance within that so why would you keep going on that so then with pacey as well as doing all that scoring metric we also allow it to learn from users intent so everything that gets approved in the system from a a visual or um a piece of copy builds the brand's taste profile across a full team yeah and it learns from what a user's taste is and what they like and don't like so they're and it keeps building that so the system now is going okay i know what good is and i know what bad is because users and humans are the ones telling me this is what good is and what

12:26Evan Greally: bad it's i love that i love that so just maybe let's just explore what the typical user's day is because there's some like really nice features that i i really took to like i love the ability to go in and get a summary of what's going on within your marketplace and you can tell me if this has changed since i the demo that i saw back i was literally given a like a radio show with a with a beautiful irish voice telling me what i specifically requested can we just keep the irish voice because it's great so i want to know about my marketplace and i think that was funny because

12:57Daniel Rowles: when i thought of that idea i was like oh this would be a nice little gimmick to put on but it's actually one of the most popular parts of the app so every day when you get your daily briefing you get the the podcast as well and has two speakers very much like notebook lm but it's it's it's all about your what's in your daily briefing and what matters and it kind of brings you through a whole story arc in about three minutes and people just listen to it in the car as they go into work or they sit down with their morning coffee and instead of having to read all these things that happened you can listen to it and then you can action it through that system and i think you know you come you come in the morning you see something happened pacey has either created content for you that you can then approve or change how you want or it will say oh this happened would you like to action it and work with me to build out something and whether that's a social campaign to a visual asset so within pacey we work across the team from the social media manager who has to make sure that the schedule and everything is correct so pacey allows scheduling and posting and you can put up your whole year of content if you want to i don't suggest doing that but you can't um and it's about managing that calendar as well so even within the calendar you can add events say you have a new sale coming up in september pacey now knows this is on the calendar and two weeks in advance or whenever you want it to tell you it will say oh the sale is coming up you need to write content and that content then can be graphic designers working in there because with pacey we don't generate flat files like chat details or any deep things i love this i love this

14:33Evan Greally: because this is this for me was really game-changing because i've seen lots of systems that will output graphics that you can post up onto your socials and stuff, but you can't get at them. So if a word's wrong or the image is wrong or the colors are slightly off or whatever it is, you're a bit stuck with this. What you showed me gives you like a layered file with everything edited.

14:51Daniel Rowles: Exactly, yeah. And with the layered file, it's also pushing back against the solar slop and AI slop taking that because, you know, you can go in and edit it and then that's an owned file that you can then save as a template. You can then hand that off to anyone to change copy or visual to add different logos. And as well with that, because we work with multi-outlet brands, they might have five different restaurants in a resort. And each restaurant has its own logo and its own colors and its own voice. And Pacey takes all that into consideration when it's designing something so that it's like, okay, I'm looking at these colors and this font and this logo now.

15:25Daniel Rowles: And it will generate that piece of content relevant to that piece, that outlet, whether that's a steak restaurant or a gin, for example. and the layered file then also allows for people to use their own photography so a lot of our brands we work would have their own really good photography and instead of having to generate an image of a family playing on a beach using chat for d or nano banana too they get to use their own photography but then with the graphics layered on top but then it also pushes a past social media where pacey can generate your 48 sheet billboard if you want it because we can do that high-res

15:59Evan Greally: content amazing like who would you say that the tool is actually for like who've you aimed it at like is it for a particular size of organization or is there a particular fit with particular types

16:09Daniel Rowles: of businesses do you find so it's it's definitely within in-house marketing teams but the way it works is less about the the team number and more about the complexity the more data you have the better pace he actually performs um so brands with a lot of multi outlets and reviews and everything perform best. Amazing. Why do you think marketers should be excited about this? Nobody got into marketing to copy information between six different tools. You know, people got in to understand customers, have ideas, make good work. And that's the job that Pacey's giving back. And we're kind of trying to shift the role from operating this process and this workflow that a lot of teams that are just stuck in to allowing people to now direct it, you know.

16:53Daniel Rowles: So they're still deciding what's right for the brand. They still approve what ships but they're now given the time to actually think is this right for my brand not oh i need to get these six social media posts out because if i don't get them out i'm going to get

17:06Evan Greally: shouted at by my cmo it's worse than that it's like that that that question you get from a very

17:10Daniel Rowles: young child what did you do at work today dad exactly yeah oh yeah i i cut and pasted a load

17:15Evan Greally: of content into multiple platforms and pushed it out for people to ignore it people's ability is

17:21Daniel Rowles: so much more than this and we just we've completely lost sight of that because there's more i think you know there's a 90 increase in marketing content since 2017 and 90 increase like that's and now you've got like chat video ads to worry about and how how do i my aeo and geo search and seo is still kind of there but in the back like a little bit scared that it won't be the week or something but the point is we've lost sight of what it means to build a brand building a brand isn't that I put out a post every hour on every different platform. It's about the right content, the right message at the right time.

17:57Daniel Rowles: So I'm talking to my audience and they want what I'm selling. And that's the thing that we're trying to get Pacey to give back to marketing themes.

18:06Evan Greally: I know what a lot of our listeners would be thinking. Well, can I just build this as a collection of skills in AI? That was how I approached this actually when you gave me the demo because I was adamant I wasn't going to like this. And the more you showed me, I just thought, no.

18:18Daniel Rowles: I think the thing with Pacey isn't that you can definitely build versions and prototypes of what we're building. But where Pacey differentiates itself is the filtering and the understanding of what good data is and what data I need to see as a brand. Because when a brand comes in and onboard, we don't query every data source the same for every brand. This is how I, for this random hotel brand in Spain, query Reddit to understand that brand better and how their audience is talking about them already. And it's completely different for a set of dentists or completely different. So that filtering is the bit that's highly complex.

18:59Daniel Rowles: and then the layered file stuff as well is you know it's understanding how you do like negative space and design and giving it all the design tools that you would have in photoshop to be allowed to create the kind of content that we need to create that's really high quality you know we're also now introducing next week video end-to-end and these are editable videos as well like motion graphic videos so it's like our pacey graphic system it's also now this video system that isn't just you know google omni generating a marketing video it's again layered video file where you can go in and change all the copy and people aren't using it to generate ai video they're actually uploading their own video clips it's editing it together for them the understanding that we get from building out these systems is that it's incredibly difficult to keep an AI model on track and the rules that you put in place are the actual specialty and the constraints that you put on the system and the know-how and the constant reviewing and evaluates on everything that we build is the bit that's just very complex to recreate.

20:06Evan Greally: When we were chatting about this you gave me a really good analogy that you know very often the way marketers are using specific AI tools is like as like one instrument but with Pacey what you've set about creating is a whole orchestra that can play like whole pieces end to end and for me that was just a really beautiful image and i think it summarizes it quite well about curating the ai models that we use for

20:26Daniel Rowles: everything so we're not beholden to any one platform in that sense that we use gpt 5.5 for certain things and claude uh opus 4.8 for other things and gemini for different things so like it's about understanding what each model is best at and giving people access to that but without all the fluff of oh my god a new model released yesterday i need to get on it right away it's like someone should be to worry about that like don't worry pacey's got that covered we'll give

20:52Evan Greally: you the best model for the best job when you need it and it's at your fingertips and i think you know you're right in ever increasing tech stacks that we use these layers that get in the way that previously we had brand teams and they knew the brand inside out and they they kept the orchestra like on on track playing the right the right tunes right but actually increasingly as all these other tools have come in and they've come in so rapidly i think actually you end up with quite a disjointed and jar like whole experience and actually that's that's not best practice but you know people are doing it anyway because it's faster and cheaper yeah i mean it's about keeping

21:25Daniel Rowles: that to keep on the orchestra thing but the same note and the same melody throughout the whole sequence from this happened in my market last night to this piece of content just went out and it's the same context that it always has but the orchestra around that is the graphic design agent is the copywriter agent is the creative director and the strategist and they're all working together to to create the best content for a brand and just tell me what what sort of things have you

21:49Evan Greally: adjusted within the product so far because i'm fascinated with the whole kind of build story

21:53Daniel Rowles: that you must have gone through with this yeah i think the the biggest change for me was like i've built the best marketing tool in the world here's a login go and use it and everyone went okay cool but the onboarding was the the biggest thing that we really learned that we need to slow down on because it is a different way of working it's the it's the new way of doing marketing when when we come into systems like Paisley and it is about building trust with teams so that first week is now a lot slower than we initially would have planned for but the onboarding is the most important part because we build trust with teams we bring them through the process we show them how the data flow works where the data is coming from how pacey graphic makes decisions how our video service makes its decisions and then just showing them that the system isn't hallucinating was the biggest thing you know like when you're onboarding it's like oh it said this really cool thing here but like is that true yeah and then seeing the six sources and clicking into each one and not only clicking into each one but it like hyper leads to the actual section where pulled that content out of to show you exactly why Paisley came to that conclusion.

22:58Evan Greally: And where do you think this will be three years out from now?

23:01Daniel Rowles: I would love to be the most agentic marketing tool. Basically what Figma became for designers. I want Paisley to be for marketing. And loved is a very deliberate word because, you know, most B2B software dies in just being unopened. Yes, people are paying for it, but they don't renew and they just kind of go, well, it was cool for a week and then it's gone. So we're trying to solve those three problems off the data problem because we're drowning in everything. The tool problems, too many tools, too many different LLMs coming out every week, too many different design tools. So it's about cutting that stack down to Pacey can do it all for me.

23:41Daniel Rowles: That's my ultimate goal is that we're keep true to our honest silence. Honest silence is also not just about data. It's also honest silence of the tool not becoming this huge thing. the things you need to do code marketing. And that's our goal.

23:56Evan Greally: So Evan, thank you so much for taking the time to tell us all about Pacey. Guys, I hope you've all really enjoyed this. Hope you're inspired to go and take a look. If you want to find out more, where should we go? What should we do?

24:05Daniel Rowles: If you sell a service, go to pacey.ai and book a demo call. We'd love to hear from you and show you the platform. Or you can find myself, Evan Greeley, on LinkedIn. We post weekly updates of what's happening with Pacey and the features that we're building. Or you can also just reach out to me on LinkedIn. I'm always open for a chat. Amazing. well Evan we wish you the best of luck for more episodes resources to leave a review

24:28Ciaran Rogers: or to get in contact go to targetinternet.com forward slash podcast