Become a Creator today!Start creating today - Share your story with the world!
Start for free
00:00:00
00:00:01
Mind the Model: Ep 012 Catherine Rushton
42 Plays16 days ago
People First, AI Second — with Catherine Rushton
Episode Overview:
Nathan Guerra and Emmalee Crellin are back to dive into the intersection of human intuition and machine learning. In this episode, they sit down with Catherine Rushton (affectionately known as "Crush"), the Chief Strategy Officer and Partner at This is Flow. Crush shares how her agency uses AI to empower their team and why she believes human intuition remains the ultimate competitive advantage.
The TL;DR
- The AI Breakup: Both Em and Crush discuss their recent "divorces" from ChatGPT in favour of Claude, citing better tone and writing capabilities — though they note Claude's high token consumption.
- The "Happy People Project": Rather than a technology-first approach, This is Flow uses a people-first philosophy. They empower junior staff to audit their own workflows and identify "unhappy" manual tasks for AI automation — speeding up the menial so humans can slow down on what matters.
About Our Guest
Catherine Rushton is an award-winning strategist with over 15 years of experience driving high-impact communications for global brands like Dove, Chanel, and Volvo. Now a partner at This is Flow, she focuses on building elite, high-performing teams that can outperform algorithms by leveraging human intuition.
📚 Resources & Links from this Episode
- Crush's Prompting Tip: "Outsource" your prompting by asking the AI itself to write the best prompt based on your verbalised thoughts. Take your AI for a walk!
- Speech vs. Typing Efficiency: A Stanford University study found that speech recognition is over 3x faster than typing on mobile keyboards for English users. Level up your dictation game with Wispr Flow: https://wisprflow.ai/
- Switching to Claude?: Divorcing ChatGPT and flirting with Claude? Import your context and memory from other LLMs here: https://claude.com/import-memory
🎧 Enjoying Mind the Model?
If you liked this episode, follow, rate, and share Mind the Model with your fellow AI-curious friends. It really helps us grow and keep these conversations going!
💌 Got thoughts or ideas for future episodes?
We'd love to hear from you!
Drop us a message at mindthemodelpod@gmail.com. Your questions and suggestions help shape the conversations we have on the show.
Recommended
Transcript
Introduction to 'Mind the Model' Podcast
00:00:11
Speaker
Hello and welcome to Mind the Model. I'm Nathan Guerra. And as always, I'm joined by my co-host Emily Carellin. Em, how's it going? Hello, it's going well, Nate. how are you? Good. Adjusting to life with a new job new phone. he's just now My recording device is new, so everything seems to be working.
00:00:28
Speaker
um Yeah, it's um yeah good though. What's up? what's shaing Very nice. Yes, yes,
Emily's AI Tool Transition
00:00:33
Speaker
indeed. um What have you been AI-ing recently, Emily? Well, it's funny you asked, talking about something new.
00:00:40
Speaker
I had um a bit of a breakup recently. Oh, I'm sorry to hear that. I divorced ChatGPT. Ooh. I know. Yeah. yeah I know. After, after a bit of a year of pain for the pro subscription, I had decided, you know what?
00:00:57
Speaker
I am sick of the sea of sameness that's coming out at me. No matter. I've really tried with different variations of prompting, tried to push it in new levels. And it was just, I don't know. I got fed up.
00:01:11
Speaker
It also probably doesn't help that Sam Altman is not exactly my favorite person. Where have you gone? I'm now paying for Claude. Um, yes, I, i haven't dabbled in Claude co-work yet. I am very keen to, but I'm going to be doing that on my personal laptop, um, and adhering to my company's AI policies. Very good.
00:01:31
Speaker
Initial reports, it's really good. Really, really good with writing and understanding the tone. I was helping my sister-in-law last night with her resume. And so going back and forth and it was a short, sharp and like would rewrite something and have it be clear on like the first go rather than kind of pushing it a couple of different ways and not rewriting things just to rewrite them, which I found was quite quite often the case with chat tbt so i'm i'm excited however the only thing i have noticed i was doing some like really big requests um using the newest ah latest and greatest opus model from claude and i had chewed through my daily tokens in like an hour
00:02:18
Speaker
And I remember messaging you and I'm like, oh, God is a very hungry boy. um Because I didn't think it was that big of a, you know, task that I was using. was taking different locations and asking it to model it out. And on a Google map, getting a.kml file, like all this stuff. And I was like, whoa, whoa, whoa. And then all of a sudden it's like, nope.
00:02:37
Speaker
Take a pause. of like, oh. e lame Yeah, um I've heard repeatedly from many people that would echo your experience that Claude is very good from a language perspective, but also a very, very hungry token user. and And maybe the allocation of tokens is less. I never felt like I was running out of tokens in Gemini. Well, it's funny because I actually asked Claude that same thing because I tried to upload like a very heavy document and it said like, sorry, no, doesn't, does not compute.
00:03:06
Speaker
And so i was like, okay, well, Claude, if you can't do it, who can? And it did come back and say Gemini Pro actually has a larger token context window. So maybe that's why you weren't and capping out as fast.
YouTube as a Resource for AI Solutions
00:03:17
Speaker
um Yes, well, I've been dabbling quite a lot with m three six five which has been microsoft yes yeah quite a, um i been it's been a very useful tool as I've onboarded into a company that is a M365 complete shop. Everything that we have in the business is in there and it's been really powerful at finding the right information.
00:03:39
Speaker
um And very helpful as I've like now moved to Windows and I'm trying to figure out keyboard commands and I'm struggling with Outlook and I'm doing all that stuff. And so it's been a very just constant companion for me recently. um I will continue to report back, but it's it's definitely become much more part of my life. Well, you know, I have a bit of a top tip for our listeners this week. If you have ah a weird thing you're trying to do or trying to unlock some new gain, I'll be going back to ask like why Claude isn't giving me exactly what I need to. Why am I chewing through tokens so quickly? But yes, YouTube is our friend here in this world in crazy times of AI.
00:04:13
Speaker
I could talk about this stuff all day, but I really do want to get to our guests today. Long time coming.
Guest Appearance: Catherine 'Crush' Rushton
00:04:20
Speaker
Long time coming. We've been, this has been our most re I think maybe our, maybe our most or second most rescheduled podcast. I mean, Roger Dunn is really up there as well. We'll lock in and eventually. Eventually we'll get Dunny.
00:04:32
Speaker
um But Catherine Rushton and I go way back. We have ah formed a club together. and She's also an award-winning strategist. She's the chief strategy officer and partner at This Is Flow.
00:04:45
Speaker
um She's been working for 15 years. at ah she's Her CV says she has driven high impact communications for brands, including Dove, Chanel, I assume it's not channel, Chanel, Tourism New Zealand, and Volvo. ah She treats human intuition as the ultimate competitive advantage over machine learning.
00:05:04
Speaker
Well, I want to get into that. That's what I really want to get into today. And it's focused on building elite high-performing teams that outperform the algorithm. I mean, this is like this is like candy for me. I love this see this this intro here.
00:05:18
Speaker
Anyway, that's enough of the intro. i Crush, welcome to Mind the Model. Welcome. Thank you. As you said, a very, very long time coming. i would like to confirm for the listeners, it wasn't just me rescheduling, but I think we did a bit of a dance. But I think delayed gratification, hopefully for all of us, and hopefully it comes through. Yes, absolutely. 100%. Crush, welcome. It is an absolute pleasure having you on the podcast. And you are very fondly held my heart as one of the first long ah lunches I had in media.
00:05:53
Speaker
Long, long time ago in Surrey Hills. So yes, I remember you fondly as a strategist at Mindshare and it has just been amazing to watch what you've done with this as well. Probably making me blush, ah literally blush. So thank you so much, Em. um Obviously, both of you, I have been so excited to join and chat about today because I know organically when we have caught up in person, we kind of can't help ourselves. Yeah. But just to run into a couple of the quick fire intro questions, Krash, what is your favorite AI model and what's the last thing that you use it for?
00:06:26
Speaker
well, listening to your intro, I recently also broke up with ChatGBT and learned through a very helpful YouTube channel how to export all my data. Have you done that? I did. It took almost, it took like 24 hours to download all my contacts. I know, right. And I've just done it for Gemini too.
00:06:42
Speaker
um and put it all into Claude so I'm still dabbling obviously with Gemini but I have officially broken up with ChatGPT transferred my data and yes my um there's a joke in our household that we should have a little alarm to remind us when our tokens are back up again because my husband I think almost cries every night when he runs out of tokens so yes we've got a very nerdy household here but yeah I am listening to your intro I was I absolutely feel you. And what did you last use it for? I'm challenging myself on some of the answers I may give today. so some some, interview prep.
00:07:15
Speaker
I feel like from your funny laugh that that wasn't a very interesting answer. I often use it for the, those kinds of things as well. I find it's great whenever you're going into a situation where you're not really, you want to prep for it. Like AI is amazing prep tool. And stretch your thinking as well. Well, that's the main thing, right? is um who I mean, who knows what I'm actually going to add today, but I guess it's more in like the process of thinking of, um I like that it stretches me when I don't have a lot of time.
00:07:43
Speaker
So if I know I've got an hour to dedicate to this, but I really want to stretch my thinking, then it's a good little thought part. um And do you have a prompting methodology or approach that you use? Oh,
00:07:57
Speaker
I would say most recently, one of my favorite things to do, particularly with Gemini, is actually ask it to write the prompt for me. So so I kind of verbalize, literally speak into it and say, these are all my crazy thoughts.
00:08:10
Speaker
I need to achieve this. Now tell me what my prompt to you should be. And I've even started outsourcing the prompting, but i've I've learned a lot that way as well. I love that. That reverse engineering of prompting. The the ultimate outsource. i don't even want to have to think about how i think. So if you can just do that for me, that'd be great. we are we're so stoked to have you here. so I wanted to get into a question about Vysis Flow.
AI Strategy at This Is Flow
00:08:36
Speaker
What makes Vysis Flow use of AI special and different from other indies as well as other holdcos? i was actually at the I am double event last week um and Margie Reed and Katie Rick Smith were on the panel. And I actually loved what they both said and better than I can say it myself is they looked out to the audience and said, you have the both financial freedom, but also you're at the arguably smaller scale where you can experiment more. And the kind of risks of those experiments are arguably smaller So you can push the boundaries and you know, I think of an example, um nearly three years ago now, this is Flo launched the nine day fortnight, which we still do. um and as a kind of people-based initiative, we did something that would have been very difficult for the holdco agencies to do with, you know, thousands and thousands of people. And because at the time we only had about 20, but now we've got about, we've got 52. It means we can roll out these initiatives faster,
00:09:37
Speaker
We can kind of iterate on them arguably faster as well. So I think some of those things naturally give us a starting point. but That's about your business. What about AI? Like how does, what, how is it that you guys use AI differently?
00:09:51
Speaker
Good challenge. um So I would say i can't speak for the inner workings of other agencies, but at Flow, We hold on to the kind of same philosophy that we started with, which is two of our values, stay human and keep evolving, which are pretty basic on the surface level, but they drive so much of what we do. So when you apply stay human to AI, it seems counterintuitive, doesn't it? But I think it's, we have a people first approach. So the first thing there is for both our clients and our talent is we start by asking them, what are they nervous about? What are they scared of? What are the problems they're seeing? And there's ah initiatives come out of that.
00:10:27
Speaker
So um we've been on a journey for about 24 months now with this, and I would say it's quite iterative. So I don't have kind of like a one big name for it of how we use it, but yeah,
00:10:38
Speaker
the um The way that we like to call it in the business is the happy people project, which I think already gives you an indication of our approach. So rather than technology first, we're people first. So a very basic example of that is how we approached our audit of automation was we got the people who are doing that grunt work to do the audit themselves. So we said, we want you to be the happiest people. What are the things that make you unhappy in your day-to-day work? And that's our approach to our automation audit, rather than senior leadership coming in, observing, reviewing, and telling them what to do.
00:11:14
Speaker
They present back to us and say what they want to do. An example of that, to be more specific, because I'm knowing you, Nate, you'll be like, what specifically does that mean though? um Take report writing, for example. We are optimizing campaigns regularly and feeding back to clients. And yes, we have experimented with kind of AI generated report writing, but we really want our teams to practice writing that commentary. So they can use AI to kind of help and um edit and kind of have that thought partner. But the point there is it's obviously specific
00:11:52
Speaker
naturally sped up the process, but I think it's also helped that learning. And of course, finally, the quality for clients, because they want rather than just the what that's happening, but also the why behind it. So when a yeah you know junior member of the team is busy optimizing the platforms themselves, sometimes they may rush the commentary, which is completely understandable. So this kind of helps balance that out a little Who in the team would you say is getting the most value from AI? The junior levels who are quite busy with the work that can be pushed to automation or the senior leaders who are now using it to scale the strategy?
00:12:30
Speaker
From an automation perspective, you would definitely say that that kind of replacing of manual tasks is a benefit, but that's not purely kind of generative AI and the benefits there. And I would say the kind of positive, particularly because I look after the strategy and data team on a more day-to-day level, um I would say the biggest transformation I've noticed is from a data and analytics perspective. So, And particularly that comparison to say a whole code with that big scale. um You know, at WPP Media, you had tens, like dozens of people and analysts working on your data and analytics because they had to do more manual either coding or kind of crunching the numbers. Whereas now, Kate, shout out to Kate, our data lead and our analysts in my team,
00:13:12
Speaker
It means they're able to use Claude, you know, and actually code as well as learn new code and new ways of modeling. um That would never have been possible before. Plus AI, obviously, and various programs means the number crunching is just accelerated from a time perspective. So we can set it. Overnight, it runs, we come back in in the morning, we're ready to go. So there's more scrappy, you know democratization of technology means that as a 50 person agency based in Sydney, little Sydney, Australia means we're able to really compete on that global stage for our from our sophistication of data analysis. has Has AI impacted the This Is Flow pitch?
AI's Role in Industry Understanding
00:13:56
Speaker
I guess either the process of pitching or or the product that you're pitching? Our ability to research new industries and new clients quickly and feel prepared to walk in has been a complete game changer and is probably why for the size that we are we can be as productive as we are so that may sound so obvious but it particularly you know on a human level for the senior leadership like myself I'm jumping in and out of different meetings every single day and I don't want to lose the quality of the contribution I give my clients whereas this means that I can kind of get that quick up to speed view of a category.
00:14:35
Speaker
So it's accelerating, your ability to understand clients. Um, anything else it's doing from, uh, uh, I guess, again, the product that you're presenting to a client.
00:14:45
Speaker
Yes. So on a very like practical level, and I guess in the context of what you guys are always kind of getting to the bottom of those practical tips and tricks, I would say, um, I continue to be amazed at the quality of design and ah in terms of the physical product we're able to present to clients. um so you know, shout out to tools like obviously Canva, ah but also just an ability to approach video, you know, video editing, voice recording, et cetera, I think has really, know,
00:15:19
Speaker
It enables indies to compete with perhaps the big resources of the global agencies who have, you know, huge design teams, marketing teams means we're able to produce content and kind of get it out quite quickly. i think that also is not just for new clients, but also current clients, you know, I'm able to get the thoughts from my head out into clients in a kind of more sophisticated um and well-designed, easy to read or consume way. And I think that's been a huge game changer. I'd say even in the last kind of three to six months, the ability to do that has um really been a step change. so
00:15:56
Speaker
Yes, and definite efficiency gains it sounds like. Yeah, absolutely. and Okay, I want to throw a spicy one to
Addressing AI Job Concerns
00:16:03
Speaker
you now, Crush. Oh, yes. There is heavy, heavy industry chatter about AI will replace your jobs. And especially when it comes to agencies, when there's so many people who work on that front end lines, how are you actively navigating your staff's fears around ai replacing them?
00:16:21
Speaker
Yes, this definitely is a spicy question. As I mentioned before, our approach is people first. And I would say that the other benefit of us as an independent agency, you know, I'm one of the partners as well. So this is very top of mind of looking after our people. But we also, I guess, have the freedom to also invest in what we feel is right. So our approach at at the moment is definitely not about um cutting jobs, but it's about a better utilisation of that kind of human capital, as some people in the industry are referring to it as. But that people power, how can we better redeploy it as opposed to cutting jobs? um
00:17:03
Speaker
I think, you know, even I saw in the news this week was a Wise Tech Global, which is kind of one of the Australian darlings of the developer space, has just cut, I think, did I say 20% of their workforce or something? um So i think with headlines like that, it is really sending shockwaves through the industry. But I would say that is one of the benefits of working for a kind of smaller independently run business is that we have the freedom to kind of redeploy that resource as we go. But I also, um it would be naive of me to think that they're not tough decisions you're making. And I think that's probably one of the biggest things I've learnt in the shift from being um kind of an employee to a business owner is it genuinely, when people say what keeps you up at night, um It's always been a concern. When we look out onto the floor, you've got 51 people that what you do every day is making sure that they have jobs and they can pay for their mortgages and their rent and things like that. So I think it's always been a responsibility for us, people first. So when it comes to AI...
00:18:10
Speaker
It's how are we best redeploying their time to get the most value for the business and for our clients. If I was going to boil that down to a sentence, is it you're actively navigating your staff's fears or around replacement by suggesting that you are always going to be looking out for your people and ah that there are are opportunities to redeploy anybody whose job is replaced by ai When it comes to tackling that uncertainty, the first is tackling it head on and communicating as regularly as possible.
00:18:43
Speaker
Because if we're proactive in explaining to them our strategy of redeploying those resources and their kind of time spent to more high value tasks, and if we're explaining our strategy and kind of approach for business growth, then that helps reassure them that that our approach will be people first and kind of keeping their job safe as opposed to just cutting headcount to kind of equal profit. That makes sense. And I think you guys are in a bit of a unique situation as well, that you can make those quite dynamic um decisions on the fly, whereas you don't have necessarily shareholders breathing down your neck.
00:19:21
Speaker
being ah inspired and excited to grow so that we can provide those opportunities for our staff as opposed to having that mentality of, oh, how can we shave those costs in those areas as a first lever for profit growth? um Because that probably is the biggest difference. With such pressure on publicly listed companies, and I see it every day with my clients,
00:19:47
Speaker
to deliver those cost cutting measures to deliver profit on a balance sheet. um I think that's probably one of the biggest differences you mentioned that freedom that we have. Actually, I'm kind of curious how you're seeing your clients use AI.
00:20:01
Speaker
the really great marketers I'm seeing out in market are playing with it themselves. So they're using large language models every day, listening to podcasts like this one about tips and tricks to how to prompt, because I think the reason why I guess labor on that is the ones who are experimenting and are more familiar are therefore more comfortable with trialing new things. Okay.
00:20:26
Speaker
Because, of course, the worry is that the brands that are doing or the marketers who are doing nothing and not trying new things are the ones who are going to be left behind.
00:20:39
Speaker
I use a more specific example of watching the launch of Google AI search mode. And that is quite, it was a fundamental change to search. And we are now a couple of months on seeing dramatic differences in results for brands. With now, I think I just saw the latest data that 47% of Australian searches in Google now result in zero click.
00:21:03
Speaker
you know, paid search has been completely turned on its head and around and kind of thrown out the window. So um I think this is a really great snapshot of the different ways that marketers and brands have approached this challenge. um The ones who are naturally interested in this space were the ones who either lent in or actually came to me and were like, crush, what's happening? What's this going to mean? Whereas perhaps others are unsure and a bit deer in headlights. So I guess that's where our responsibility comes in to make sure that we're guiding them on that journey. Are you asked are you telling your clients to change how they're investing in search then?
00:21:43
Speaker
Absolutely. As an example, with some of the changes that have been made, what we are seeing is that more quality traffic is coming through. Now, that kind of makes sense logically because some of that rubbish traffic that was probably being just funneled through paid ads um is not there. Also, I had a client the other day make ah a good point, which she was like, some of those kids who are researching top destinations in New Zealand for one of our travel brands, ah those kids who are kind of just bumping on for school projects are not coming through. So we're actually seeing...
00:22:17
Speaker
ah surprise, I would say, results in a more positive way through the changes that we've made. um want pivot slightly to your bread and butter, which is strategy.
00:22:28
Speaker
How have you seen the uptick of AI impacting the role of being a strategist?
AI's Impact on Strategy
00:22:34
Speaker
Yeah, this is a big one. um i would say as naturally curious minds, this has been one of the most, i I would say the most exciting thing that's happened in my whole career. Because as people who are naturally curious and hungry to know as much as we can about as many topics as possible, We have it now really at our fingertips. But I was like, yeah, as you can see, totally geeking out and nerding out. and As you should be. It's so exciting. Exactly. um
00:23:10
Speaker
i would say also for me, just starting personally at this stage of my career You know, I went from having a team of 12 strategists who I could lean on to kind of fuel my thinking and get me up to speed on everything. And then I moved to This Is Flow when I first joined Jimmy and it was just me.
00:23:27
Speaker
But fortunately, that coincided with the rise of large language models. So it made me scrappy and kind of get back on the tools. Yeah. So obviously from an efficiency perspective, but also more exciting, more excitingly, not everyone wants to talk to me about strategy all the time, but I have this kind of. not? I find it fascinating. What? What? not having that big team to bounce my thoughts and ideas off, um, but also perhaps sometimes you're self editing because you're worried that you're random train of, you know, a stream of consciousness, you may be self editing. So the great thing with large language models is, and I really encourage my teams to do this, is this kind of freedom of expression because it's when you go down the rabbit holes,
00:24:19
Speaker
is when you're going to find that brilliant. One of the options that people sometimes will use is they'll literally speak to their AI assistants. Is that something you're doing then? And and is that how you're getting that stream of consciousness out?
00:24:29
Speaker
ah Depends on the task. Also depends where I am because um when I'm in the office, it's a little more challenging, but I'll give you a really um specific example. So the other day, um it was actually the day after ChatGPT globally made the announcement of ai ads.
00:24:46
Speaker
And obviously my brain was like, need to have a point of view on this client. I had just seen a client that day and they said, crush, what does this mean for us? Um, so I'd obviously been tinkering. I'd been doing a bit of research, but I had to get my car serviced and it meant I had to take a 20 minute walk back from, from the car being serviced. So I put my headphones in, i opened up my Gemini and I started saying, I think this might be the implication. I wonder what happens here, blah blah, blah, blah, blah. As I'm walking, getting all my thoughts out. And of course, by the end of that walk, I had a really nice, neat draft of of their thing to send out to my clients sharp
00:25:30
Speaker
um you know, strategic thought leadership piece ready to go. Now, so that's obviously a really um great example of not only being challenged about what my thoughts are, getting me up to speed, but also very efficiently getting something useful for my clients out. So yeah, voice, absolutely agree.
00:25:47
Speaker
Also geeky strategy side note, obviously your brain processes information differently um and you can speak a lot faster than you can type.
00:25:57
Speaker
ah So it is a really good way to get it. Yeah. I've actually been looking into whisper flow and saying out loud, maybe I should go ahead and download it. Cause I am notorious for voice notes. Nate can attest to this. I just love talking into my phone. um you so You saw me visibly jump. um One of my friends, shout out to Lucy Ko, she's a founder of Tap the Gap. Amazing initiative, by the way, on the side note. But I was away with her on just two weekends ago and I walked into a room and I literally saw her to her computer just whispering into the side.
00:26:31
Speaker
And I was like, why are you whispering? And she's like, whisper flow. Haven't you heard? And so you and I, let's make a pact. We're going to go and investigate. We're going to go to pilot this week. And your next podcast, that can be the thing that you've been playing with. want to go back to the impact of AI on strategy.
00:26:49
Speaker
if we were going to have the same conversation in three years' time, What AI driven capability do you think would be like stock standard for any media agency today? think the one that I'm most excited about is, you know, context based media planning is not a new thing. ah Back in, you know, at the beginning of my career, I worked on Kimberly Clark, good seven years of my career spent on that. And they have this framework called moments of receptivity. And I'm still one of my all-time favorite clients, shout out to them. But moments of receptivity, right? So you use time of day, location of the audience, maybe mood, kind of are they leaning forward on a computer or are they leaning back watching TV? And you can kind of conceptually create this beautiful experience that your ad should land in.
00:27:43
Speaker
But the problem with that, and that's nothing new, that's been around for for decades, but I think what I hope, and imagine will happen is we will get more richer signals than ever before um about how a human may be receiving a message and kind of what receptivity moment they're in. So if I take a few examples. So ah if we think at the moment that 30% of Australians now use ad blockers and it's rising every single day, that tells me that a lot of Aussies don't like ads.
00:28:20
Speaker
Which is a problem for us. Existential shock. Wow. What a surprise. It's one of the highest rates in the world though. So, um but also we have huge subscriptions. So Gen Z in the recent Deloitte research, Gen Z now have on average $101 a month they spend on paid media subscriptions, which is another way of avoiding ads, right?
00:28:43
Speaker
That's on average. um So I think ahead to three years time as AI slop increases, more and more ads are out there and agent to agent kind of There's ads flying everywhere. What I would love to see is a more respectful approach to media delivery and kind of media planning. Because, for example, if everyone's wearing a watch or a kind of a ring that tells you their heart rate, how well they slept overnight, um you would argue that someone who's well-rested
00:29:16
Speaker
and kind of happy in mood, maybe more receptive to certain messages. Or if we know that someone's just been on a run or they did go to the gym this morning or didn't go to the gym this morning, maybe we have to change the tone or the speed of our ad. And I think being more respectful of people's attention, now you can do it. You can kind of like hack it together, but you're not guaranteed that it's going to deliver the experience exactly as you want. But I do think that um I can see a...
00:29:45
Speaker
a more peaceful, enjoyable ad experience in the future. i would also argue that agent to agent will force that because agent your agents will probably block you from seeing things that are not right for your mood. um So i can also see that it will kind of agentic delivery will mean that we'll probably be forced to be better.
00:30:11
Speaker
um But yeah, now I'm really extrapolating out into the future. But what do you guys think of that? That requires even more personalization, even more data, even more connection, controls and even more privacy controls. And I guess what I'm thinking you're suggesting is that That privacy is actually then handled by your agent just communicating my client wants this as opposed to you giving that information to the advertisers of the world. in On that practical note, I could imagine that it is the agent saying, I'm ready for this.
00:30:44
Speaker
you know, I'm ready for this type of ad, I'm ready for this type of moment, or my human is, as opposed to free flowing information out into the ether. But I guess in saying that, though, and if we take, say, even like, you know, first party and third party data signals that have been available for years now, we can on an an anonymized level, look at a kind of a meta level analysis of mood and things. So we could probably still start to see trends that um are being reported back, even if it's not on a one to one level. But yeah, good clarification, Nate. You're certainly not getting an anyone's heart rate today.
00:31:17
Speaker
in no I mean, if you are, I'd love to know more about that. you know, maybe that's the disclose secret is... maybe a different cut Maybe in some of the other countries we operate in. joking.
00:31:31
Speaker
it's not to be joked. Switching gears are a little bit, you're co-chair of the MFA Awards again this year, as you were last year.
AI in Judging Process
00:31:40
Speaker
um I'd love to know, like, what was the impact of AI on the awards? Both, I guess, the entries and how you, the process of judging it. I am very proud and excited that our experiment in integrating more more kind of AI, but more specifically use of large language models into the judging process, this in 2025 was ah definitely, we consider it a success. So shout out to our long running partners and um sponsors, Google. And I really, one of the things that I loved about working with them is they've really challenged us to integrate that technology into the judging process. um More specifically, we used Notebook LM. One of the benefits of Notebook LM for anyone who hasn't used it is it obviously feeds off the Gemini model.
00:32:30
Speaker
I say obviously talking to Nate, ex-Googler, but... um i It feeds off the Gemini model, but the huge benefit is it's closed in terms of its use. So it's not going to go any data you give it is not going to go back into the model. So we could be really confident as a committee that um the really important confidential information that clients and agencies have trusted us with is not going to be going back into the model. So that was the first foundational layer of our approach is safety and kind of um privacy.
00:33:01
Speaker
What we were then able to do is look at how do we make the judging kind of uphold our rigorous standards of judging. So making sure that all the entries actually met the eligibility criteria. That can be quite a cumbersome process for judging. the team that works behind the scenes. So what we were able to do is speed that up. So the first check last year it was that Notebook LM would review all of the entries and the criteria and just do that first check that it was in the right time period, it had all the right information, etc, etc, before it went to the human judges.
00:33:39
Speaker
I have to um ah emphasize that there was a human in the loop in that process. So there is no concern from any agencies that you were accidentally kicked out of of the judging.
00:33:51
Speaker
um And then I think the thing that I'm most excited about we did that was more fun was how do we make the judging experience better for our judges? So we have industry, you kind of sort that the industry titans and kind of very senior people in agencies and very senior CMOs they're volunteering their time and we want to make sure that they have the best experience possible because we know that they, we also have a bit of competition. We have a couple of awards ah that we're competing against. So we want to make sure that that's great. So what we did this year is, um for example, in the innovation category, we used Notebook LM's podcast feature. So we took all of the entries
00:34:30
Speaker
um with permission. And we made them into mini podcasts that the judges could listen to because for anyone who has judged an awards before, an industry awards like this, there are walls of text that you have to read, right? um So we turned it into a little short two to three minute conversations where it brought to life the case study and the feedback was overwhelmingly positive from judges. And have you written this up anywhere? Is it like, you know, is there an article or are you hearing other awards doing the similar sort of things? I mean, this feels very um cool and exciting to me. like it Oh, that cults my ps sidekick could be very happy to hear that. I'm sure. um But maybe it's a note for cults and I to go and write up some sort of case study. ah But also maybe we don't want to give away all of our secrets. I have i've just told everybody on a podcast as well.
00:35:15
Speaker
Exactly. All the deep dark secret. Because of course, well, one of the other benefits is that all of those podcasts can then be shared with MFA members um so that people can learn about those case studies in a different way. um On that point of awards, um one thing I am right now is award season. I'm doing a lot of the people awards like 30 under 30, Young Lions, and And the biggest shift I've seen in the last couple of years is the declaration of use of ai um So shout out to Can Young Lions. They have a fantastic little um declaration in theirs, which is quite robust and nuanced, to which I respect. So it's, yes, you have to declare if you used AI incoming or some sort of large language model in um creating your entry.
00:36:02
Speaker
So for the thought thought process, but also the design process, But the great thing is it's very um supportive and says, but that isn't a bad thing. We just want to understand how you used it. So then they have to kind of give a little description of in what ways did they use it. So I thought that's that's quite nice to see that there's a bit more of a nuanced background. approach as opposed to just did you use it or not? We shouldn't be afraid of disclosing the fact that we use AI. I'm the first one to say like I use it for everything and it should be integrated into our daily lives because as you just laid out so clearly like the efficiencies we see
00:36:42
Speaker
um how it can stretch our brain in new directions. If you're you know quite under the pump and you don't have the headcount of 12 people working underneath you, like how it can get you up to speed faster to present to clients in meetings. We've seen so many benefits. And and yeah, it makes sense to wear that badge proudly. Absolutely. And I don't know about you guys, but as I'm kind of getting on in you is getting on in years in my career as well, that you...
00:37:08
Speaker
you're stretched and you're creating those new neural pathways because you're learning new skills every single day and you're approached with new technology that you've never used before. So like me battling with the microphone set up today, um it stretches me and, you know, keeps my brain um keep my brain working. So why shouldn't we be encouraging that and sharing those examples? Because it doesn't come with a user manual. So sharing stories like this and amazing podcasts like this where you're sharing tips is how we're couple of closing questions if you've got the time.
00:37:39
Speaker
oh yes. Rapid fire. Do you consider yourself to be an AI optimist, realist, or pessimist? Realist. i I would say I was naively, ah like I'm an, as you can probably tell, i am naturally an optimist and like to see the good things. Um,
00:37:58
Speaker
But over time I've realized we should proceed with caution. And even if it's on a day-to-day level of AI literacy of really understanding the implications of using tools like LLM. Claude, for example, has just, you know, countless examples of revealed to be searching things that it shouldn't be despite being told not to. So I think it's also our responsibility to be a realist. You're our first realist. so and um And a well-reasoned realist it was.
00:38:27
Speaker
I think I agree with with you, Krush. It's the realist aspect. it just There is a responsibility on us you know to to use it safely, to ensure that you know we play within the guardrails, and then to understand how what else we need to develop so that we aren't you know all replaced by robots one day, as Nate likes to call our robot overlords.
00:38:49
Speaker
Exactly. i mean, we may already have been replaced and this is just like an experiment they're laughing at us at, but um that's a whole nother. We don't want to go down. um But, you know, I think it is important to discuss that nuance because also i guess it's partly my role as an agent for my clients is I have to take on that responsibility. Also, probably since I've become an independent business owner,
00:39:16
Speaker
You've also got to understand the risks that are involved as well. Last question for you, Crush, before you go Finish this sentence. In three years' time, AI will be blank for agencies.
00:39:28
Speaker
AI will be invisible to agencies. So we'll be so into, i mean, it it already is fast becoming, but I actually have to shout out to Katie Riggs-Smith, her beautiful thought pieces on this, because um as the technology becomes more intuitive and, you know, more integrated in everything we do, it becomes invisible. So we kind of forget that we're using it. So yeah it will become invisible to agencies.
00:39:54
Speaker
I love that. What a great way to end.
00:39:58
Speaker
Crush Catherine Rushton. that's not rush i always want to do that's I sound very royal when you say that. That was fantastic. Thank you so much.
00:40:10
Speaker
Thank you for guiding me through the experience.
00:40:17
Speaker
Nate, how did you find that interview with Crush? Oh, I mean, isn't she the best? Such a smart operator. I love what she is doing at Business Flow, especially that staying human and keep evolving that people first approach.
00:40:32
Speaker
Yeah, I definitely think they've cracked the code. Great business, and they've been winning lots of awards and lots of business. So ah yeah, kudos to them. um What did you take away from the chat? What's interesting, she brought in that little bit of insight from the independent Independence Day that was recently in Melbourne about ah Indies having a unique responsibility to play at the edges. And I think that translates really well into how she was speaking about her experience with AI and, you know, upskilling the team.
00:41:02
Speaker
as well as you know playing with it herself and more of a personal use. and What about you? What did you take away from it? i was really impressed with, and she talked about even just the small tips around using voice.
00:41:15
Speaker
I thought that was just a really great tip and a different way of thinking about AI. We um often are just used to writing with it, but you know using your voice, we communicate differently. I think powerful. Just simple tips like that I thought were great.
00:41:28
Speaker
I so loved her optimism about, you know, better advertising is just over the horizon. i mean, yeah it's what we all hope for. And I think for her being more of a strategist viewpoint, I can tell that she's, you know, really keen to see that come to life, but we'll see.
00:41:44
Speaker
think that the work that she's doing with Colts on kind of changing the MFA process of judging and bringing AI into that, um use of notebook LM, um all of those kind of things I thought was very forward-looking. The and NFA Awards really does highlight like the best of the best and how they, um and even Canline that she called out that, you know, disclosure of AI incorporated into the award submissions.
00:42:13
Speaker
Great episode. loved Loved having crush on finally. We've been working on this one for quite a while. And to all of our listeners, thank you so much for your support. be sure Be sure to like, share, and subscribe. Email any questions you may have about the show to mindthemodelpod at gmail.com. And remember, the intelligence might be artificial, but the wins are real.
00:42:37
Speaker
See you next time. Bye.
00:42:44
Speaker
This episode was brought to you by Testchimp. Testchimp is a web developer's best friend. Testchimp uses deep runtime context to automate your testing workflows, generate scenarios, write scripts, and fix bugs all in one place.
00:43:00
Speaker
Stop wasting time and start shipping quality code faster. Bugs suck, but finding bugs doesn't have to with Testchimp.