Latest / Shiny New Object - a Marketing Podcast / Episode 305 / Anastasia Leng (CEO & Founder, CreativeX) & Thorsten Schapmann (Global Media Director, Beiersdorf)
Transcript
- 0:00Anastasia Leng: Think about how you can create, essentially,
- 0:02curiosity pipeline or process for yourself that helps you
- 0:06figure out and make sure that you can challenge your own
- 0:08biases as you're driving to an answer, and have that be a
- 0:12repeatable system you follow when you're working on any
- 0:15problem.
- 0:16Thorsten Schapmann: Sell the so what? So how do you get really
- 0:19to the essence, and how do you tell the story that will bring
- 0:23impact, insights and change something or moves the needle
- 0:27out of the data?
- 0:34Speaker: Are you a fitness loving foodie? Then defy snacks
- 0:37is for you. Defy offers real high quality chocolate you love,
- 0:40combined with functional ingredients, giving you as much
- 0:42as 26 grammes of protein per bag. So you can be both a little
- 0:46bit bad and a little bit good at the same time as a female owned
- 0:50brand, defy dedicates a portion of its profits to other women
- 0:53owned businesses currently sold in the US. You can find us on
- 0:56defy with an i snacks.com and follow us on social at defy
- 1:01Tom Ollerton: Hello and welcome to the shiny new object podcast.
- 1:01snacks.
- 1:10My name is Tom Ollerton. I'm the founder of automated creative,
- 1:13the creative effectiveness ad tech platform, and this is a
- 1:16podcast about the future of data driven marketing. Every week or
- 1:19so I have the pleasure and the privilege of speaking to one of
- 1:22our industry's leaders. And this week is sort of different,
- 1:24because I'm speaking to two and it's also quite unusual, because
- 1:28this is my last day in Cannes, and I think everyone's very
- 1:32proud of that who's here, but everyone who isn't in Cannes
- 1:34doesn't care. But just for context, I'm sat here on the
- 1:37roof of creative X's play villa. Call it villa. It's very nice.
- 1:41I've been accommodated by those lovely people for the last week
- 1:45or so. So thank you for that. But I'm I'm sat here with two
- 1:50special guests. I'm sat here with Thorsten Schapmann, who is
- 1:53associated Vice President of Global Media at Beiersdorf, and
- 1:57Anastasia Leng, who is CEO and founder of creative x, so it's
- 2:01double trouble. So guys, can you introduce yourself to the
- 2:04audience for someone who may not know who you are and what you
- 2:08do?
- 2:08Thorsten Schapmann: Yeah, so I'm Thorsten. I am since 20
- 2:12something years in the business of digital. I started on
- 2:16corporate side a couple of years on digital agency side, and
- 2:21since 10 years, I am on Beiersdorf side with different
- 2:24roles on local, regional and global levels, since couple of
- 2:28years now.
- 2:28Tom Ollerton: Fantastic.
- 2:29Anastasia Leng: Chivalry is dead here on this podcast. But I'm
- 2:32Anastasia Leng. I'm the founder and CEO of CreativeX. This is my
- 2:36second startup. Before that, I was at Google, also working as
- 2:40both a product marketing manager and marketing manager. Have been
- 2:43the CEO of this company for the last 10 years, and am balancing
- 2:48on how to be a founder, CEO and the CEO of a fast, growing
- 2:51company.
- 2:51Tom Ollerton: So we'll, we'll start with the first question
- 2:55for you, Anastasia, so the question you guys have chosen to
- 2:58both answer the same one is, in the last five years, what new
- 3:03belief or behaviour has changed your work life for the better.
- 3:07Anastasia Leng: So one of the things I've observed about
- 3:09myself is very few things actually stress me out at work,
- 3:13except for one, and the number one thing that stresses me out
- 3:17at work is people, people issues. Whenever those things
- 3:20happen, it's the one. It will, quite literally, keep me up at
- 3:24night. And one bit of advice I got a few years ago from a CEO
- 3:27who was running a much bigger company is he said after product
- 3:31market fit, which is the stage where you know that people want
- 3:35your product, they successfully use it to achieve a goal. He
- 3:38said after product market fit, any problem you have in the
- 3:42company is a factor of not having the right person working
- 3:45on that problem. And I think that belief and that succinct
- 3:49point of view has given me the courage to make the tough people
- 3:53decisions that not only improved my mental and emotional health,
- 3:56but actually ended up really improving the way the business worked.
- 4:00Thorsten Schapmann: So we have the same question. I'm already
- 4:02very much inspired from your answer. Mine is much easier.
- 4:06What stresses me out is having just too many obligations,
- 4:10saying too often yes to things. So I definitely have much more
- 4:15nos now, like although, having said that, I'm sitting now here,
- 4:19who sent me just an invite two days ago, so I didn't say no,
- 4:23but this is something I think you need to learn to say no and
- 4:27be still polite, because otherwise you are drawn to
- 4:30meetings, drawn in contacts and drawn in topics, and that leads
- 4:35you away from the focus on what's really important.
- 4:37Tom Ollerton: I find that absolutely terrifying, because
- 4:41you're a mega, super senior global vice president of media
- 4:44now, and I really struggle with saying no. I love saying yes to
- 4:48everything. And I would have thought by the time you get to
- 4:50your level of seniority, you've learned to say no, but you
- 4:52haven't, so maybe actually saying yes isn't such a bad
- 4:55thing.
- 4:55Thorsten Schapmann: Exactly. So sometimes you have to say yes.
- 4:58Tom Ollerton: So we're going to move on now. So to advice. So
- 5:02what bit of data driven marketing advice do you find
- 5:05yourself sharing most often, like little silver bullet of
- 5:10insight that you pass on to your teams? What do you find yourself
- 5:12saying most often when it comes to helping people be a better
- 5:15data driven marketer?
- 5:17Thorsten Schapmann: So definitely, to sell the so what?
- 5:21So bringing the storytelling into the data make it tangible
- 5:26to those stakeholders, very often top management
- 5:31stakeholders who are not so deep in the data topic, because very
- 5:37often you see data driven people coming from a data analytical
- 5:42background, and they love to analyse, and if to give them
- 5:45data, the tools they can end up in a rabbit hole of data
- 5:51analytics. So how do you get really to the essence, and how
- 5:55do you tell the story that will bring impact insights and change
- 6:00something or moves the needle out of the data.
- 6:03Anastasia Leng: So the biggest observation I've made in people
- 6:06who are on their journey to becoming a data driven marketer,
- 6:09frankly, in any profession, really, is that they try to
- 6:13optimise very much for the end result. But the problem with
- 6:17that is, in every situation, they almost have to start over
- 6:21the learning process, right? They fixate so much of what they
- 6:25need to achieve that they don't really focus on, how do I build
- 6:28a repeatable system and process for consistently getting to the
- 6:32right answer? And so my biggest advice is, think about how you
- 6:37can create essentially curiosity pipeline or process for yourself
- 6:43that helps you figure out and make sure that you can challenge
- 6:45your own biases as you're driving to an answer, and have
- 6:50that be a repeatable system you follow when you're working on
- 6:53any problem, such that you are perfecting your own system of
- 6:57working and getting to an answer you have conviction on, rather
- 7:02than just getting to that answer and hoping you got to the right
- 7:08one.
- 7:09Tom Ollerton: So now we've got to know you guys a bit better,
- 7:11we're going to talk about your shiny new object, which is AI
- 7:15search, right? And what I love about this topic is, all of five
- 7:19minutes ago, we were sitting here going, Well, we know what
- 7:22it is. We know it's going to be really big, but we don't really
- 7:25know that much about it. We're all senior at our different
- 7:28levels, so I'm really interested to have a discussion and a
- 7:32debate about this thing that we that no one really fully
- 7:35understands. So I don't know who wants to answer this, but what,
- 7:39what made you guys choose AI search or LLM search as the
- 7:43shiny new object?
- 7:44Thorsten Schapmann: Happy to start first, because in our
- 7:46business of skincare, search, as always, has been awesome. Of
- 7:50course, for me, big passion point as consumers are searching
- 7:55for solutions for the problems around skin, whatever it is. And
- 8:01of course, the big G and the Google platform is over years
- 8:04now, the machine that we learned how to optimise with organic and
- 8:09paid and so on so forth. What we see now, and we have a recent
- 8:12study thatconsumers are tending more and more into LLM solution,
- 8:19of course, chatgpt being very much in the forefront of getting
- 8:23answers to their skin problems or want advice. And with this,
- 8:28of course, now the whole, let's say, mechanic that we learned in
- 8:31the past, how to drive, or organic how you be, how to be,
- 8:35on the top list, is something that we, of course, question
- 8:39ourselves. How do those models get the information? Are we
- 8:43visible where it matters? And so that's something for me, new
- 8:49shiny, but also super exciting. It feels a bit like 20 years ago
- 8:54when the whole SEO thing started. Sometimes it's the same
- 8:58thinking you have to go to the ground. What are the sources,
- 9:01how to manage, and what kind of content, or what kind of
- 9:04solutions, etc, etc, you need to, need to provide, to really
- 9:08understand, to be the, in our case, number one skincare brand,
- 9:12being the brand that has the best answer, and then being, for
- 9:17example, a solution that someone asked her.
- 9:19Tom Ollerton: And so what have you worked out so far?
- 9:21Thorsten Schapmann: Well, actually, we are really at the
- 9:23starting point. That's it. What we looking into is
- 9:29understanding, for example, those kind of sources we have
- 9:32seen, currently, a study in Germany, that a lot of
- 9:35comparison websites where external publishers are looking
- 9:42for the best, sun care solutions, et cetera, et cetera.
- 9:46Those are the sources that some LLMs are using to provide you
- 9:50with the answer. It's not that much your brand website, for
- 9:53example, which, of course, in Google is the go to. So of
- 9:57course, it's a list of and they probably change it every time.
- 10:00It's also quite interesting now, how do they make out of this LLM
- 10:06search a business when you as an advertiser, are not really sure
- 10:11what is the algorithm change of next week or even tomorrow? So
- 10:14at Google, you have this kind of releases, and you, of course,
- 10:17never publish what is the secret behind but you have a way to
- 10:21learn over time, and you can be, let's say, sure that it's robust
- 10:24for a couple of months or even a year. In those words, we it's a
- 10:29totally new and that's where we're now starting to learn what
- 10:32is the right right sources, and where do we have to change our
- 10:35content strategy, and where do we need to be top notch to be to
- 10:40be found.
- 10:41Tom Ollerton: So Anastasia, as we someone described this to me
- 10:46once it's a dark room with, you know, we're fumbling around in
- 10:49the dark trying to work out where the where the furniture is
- 10:51so we don't trip over. I think we're all ... it feels like as
- 10:54an industry when we're in a dark room of working with AI search.
- 10:57So it almost seems that like, is it fair to say that if you've
- 11:01got good SEO, normal old SEO, if you like, then, then you will do
- 11:06well in search, because if you show up successfully in
- 11:09comparison websites, if you do have a good corporate site, if
- 11:11you're mentioned in lots of different Reddits and all the
- 11:14rest of it is that, is that the way you're seeing it? Or have
- 11:16you, do you have other Intel?
- 11:18Anastasia Leng: Well, certainly don't have other Intel. I I
- 11:21don't know if SEO is perfectly transferable to LLMs, mainly
- 11:27because I think we're going to need to look at the problem from
- 11:31from a much more structured data perspective. So I'll give you an
- 11:34example. The reason I'm really fascinated about this is I've
- 11:36seen, personally, my search pattern change really
- 11:39dramatically over the last couple of months. And I'll give
- 11:42you a skincare example. You know, I'm getting old, and I
- 11:46need to hide those wrinkles. And so the other day I was, I was,
- 11:50you know, chatting with chat GPT, and saying, what are some
- 11:55of the ingredients that are so natural and proven to help with
- 11:58wrinkle reduction? And I got collagen retinol, Thorsten, you
- 12:02tell me if that's legitimate. And and then I said, Great, I
- 12:05would like a skincare product. Money's no object that has high
- 12:11concentration, I know, very relatable has high concentration
- 12:14of retinol and collagen. And, you know, it gave me three sort
- 12:20of ranked them on their percentage of collagen retinol,
- 12:22and I bought in like five minutes, right? And so I've just
- 12:26noticed that, and as a consumer, I've gotten a lot lazier. In
- 12:30some ways I don't want to troll through tonnes of web pages, but
- 12:33so all of this is a very long way of saying that, in this
- 12:36case, maybe SEO would have helped, maybe wouldn't, but I
- 12:39wonder to what extent we need to start breaking down the
- 12:43information we have in a much more structured way to enable
- 12:46LLMs to quickly find and categorise our products in a way
- 12:50that can be discoverable.
- 12:53Tom Ollerton: But my question is, is that would whatever
- 12:56product that was, would that have even showed up in the LLM
- 12:59if the SEO wasn't good in the first place, because SEO really
- 13:02is a measure of authority. Yes, it was. Don't have the online
- 13:07authority anyway. Could the LLM have even picked it up?
- 13:10Anastasia Leng: Well, so it's a great question. I don't know.
- 13:12Well, what I do know is, because I was so interested in it, I
- 13:15really just noticed my search pattern changing. What I then
- 13:18went in is I went into Google search and typed in, you know,
- 13:23high concentration retinal collagen, like moisturiser,
- 13:27basically, and did not get any of the three products that chat
- 13:31GPT recommended to me, none, right? And, and so this tells me
- 13:36that there is some, some difference. Part of that
- 13:39difference is, you know, the way that the search quality score
- 13:41works is probably biassing towards different things. Part
- 13:44of it is obviously the first things you see are ads, right?
- 13:47And sponsored listings, where relevance is a little bit of a
- 13:50different algorithm and beast. So I don't know, but it does
- 13:55feel to me like there is something different, mainly
- 13:57because a lot of SEO was relatively unstructured data.
- 14:01Tom Ollerton: So looking forward a bit here, and I don't know,
- 14:04I'm not tight on the numbers here, but I think Chat GPT Open
- 14:07AI spent $9 billion last year to make $5 billion right? So there
- 14:14is money needs to be made somewhere. And I think I heard
- 14:17you discussing earlier, at some point they're gonna have to put
- 14:20some ads within that experience and how that's gonna play out.
- 14:24But what's gonna happen there? So if I say, I want a college,
- 14:27what was it? Collagen? Collagen and retinol, retinol, collagen
- 14:33for my ageing ginger skin. Are they gonna throw an ad at the
- 14:37top and then have the then have the generative results? So how
- 14:41do you see that playing out, Thorsten?
- 14:42Thorsten Schapmann: Well, that's exactly my my question, the way.
- 14:48I mean, first of all, I agree that the search patterns in
- 14:51general are changing. It's much more personal and you, but I
- 14:55found quite interesting that you trusted an LLM blindly. Buying
- 15:00stuff that you don't know where the sources comes from. So it's
- 15:04like, I mean, I would have an opinion, but we can take it
- 15:06offline. Amazing, as always, no, the the question is, is a fair
- 15:15one? So if I don't understand as a as a brand, what is the
- 15:20sources behind how can I be sure that? I mean, of course, I would
- 15:24stand target more, very personal such behaviour, just better. So
- 15:30at least I would have known that Anastasia is very much involved
- 15:33in a skincare problem, routine solution. So of course, that
- 15:38might be already enough for me to say, okay, here I want to
- 15:41give you a solution, even if whatever is your LLM
- 15:46optimization content is not available, it's a fast path to
- 15:49at least show a brand. But in this case, I'm not sure if this
- 15:53is, will be the solution that really brings the advertisers to
- 15:59the LLM revenue stream. So I'm, I'm still very, very curious how
- 16:06it will develop, but I can't see currently, that's, that's one of
- 16:10the, let's say, the easy base. Probably they will do it because
- 16:13it's a quick wins. But yeah, that that's something I'm not
- 16:19having had my head around.
- 16:26Tom Ollerton: This episode of the shiny new object podcast is
- 16:29brought to you in partnership with madfest, whether it's live
- 16:32in London or streamed online to the global marketing community,
- 16:36you can always expect a distinctive and daring blend of
- 16:39fast paced content, startup innovation pitches and
- 16:42unconventional entertainment from Mad fest events, you'll
- 16:45find me causing trouble on stage, recording live versions
- 16:48of this podcast and sharing a beer with the nicest and most
- 16:50influential people in marketing. Check it out at
- 16:55www.madfestlondon.com.
- 17:03So I have a dark theory to share with you, right? So bear with me
- 17:07on this. So currently, I would argue that the user interface
- 17:11for an LLM or the user experience is question based,
- 17:15right? And we are using that because that's how we have used
- 17:20old search. Let's call it old search, or whatever Google
- 17:23search, or whatever searches, however I see the power of LLMs
- 17:28to be predictive and decision making, right? So for example,
- 17:33if someone starts off with their goals, right, okay, I want to
- 17:37lose weight, I want to look great, I want to feel great. I
- 17:40want to succeed at my job, and I want to have a healthier
- 17:42relationship with the people that I love, right? That that
- 17:43could be a brief that we some people could relate to, right?
- 17:46Some people that I would like, and then if you gave that brief,
- 17:50then you wouldn't even be saying, I need to look younger.
- 17:56You wouldn't be conducting the search, because you're honestly
- 18:00the helpful analogy you gave, or the story, anecdote Was that
- 18:03you, you applied a kind of Google search attitude towards
- 18:06an LLM, whereas in the future, it may just be: Look, I've seen
- 18:10that photo that you put on Instagram Tom, your skin looks,
- 18:13you know, you're getting a few wrinkles. Actually, I've already
- 18:16bought this thing for you, and it's and it's and it's on it.
- 18:19It's on its way. I need another photo in three weeks time, and
- 18:22we'll talk about the results.
- 18:23Anastasia Leng: Yeah, perhaps I think you're absolutely right
- 18:26that I used, I used an LLM to short circuit my research
- 18:30basically right to cut through all the searching that I would
- 18:35have done. It's hard for me to imagine. Or maybe this is a hope
- 18:40that it will be an intrusive ad experience similar to what you
- 18:44have now on on Google search. I do wonder how much of the
- 18:47monetization via something like affiliate links would actually
- 18:51be more applicable, right where the LLM still recommends what it
- 18:57considers the best product based on research, in this case, based
- 19:00on, you know, a search like prompt, but if you complete the
- 19:04purchase right, it receives a portion of of the revenue
- 19:09because it it essentially sending you the traffic. Now, to
- 19:11your point of predictive and probabilistic. I mean, in some
- 19:18ways, you know, this is what Google said they could always do
- 19:20because they can see what you're searching for. They don't know
- 19:22how you feel, but they can see what you're searching for. But I
- 19:25get the data set is not too dissimilar. We do... Again,
- 19:29speaking from personal experience, I do probably share,
- 19:33share things with my ChatGPT that I haven't searched for on
- 19:37Google, because it is more conversational, right? So yes,
- 19:40it's assumed it will have other information. But what I wanted
- 19:44to suggest things to me proactively. It's hard for me to
- 19:47imagine that right now.
- 19:49Tom Ollerton: But, but you, but the word you used was time,
- 19:52right, which is, is super interesting, like you short, you
- 19:55short circuited your your search behaviour. And there's a bunch
- 19:59of things, a bunch of purchases that I will make where I don't
- 20:03want to have to make them, like, low consideration, FMCG stuff,
- 20:07you know, cleaning whatever, you know, all that stuff and that
- 20:10all takes time. Every minute I spend buying soap, I don't spend
- 20:14it with my daughter, for example. So I agree with you. I
- 20:17don't want it to send an outfit for Friday night, that would,
- 20:20that would be terrifying. But actually, yes, if it was, like,
- 20:22right skincare, like, you know, you've come back from, Cannes,
- 20:26we know you're in Cannes, you have been hammering your skin,
- 20:29and, you know, you're a ginger man that stage. So, like, here's
- 20:32the thing that you need. I, and I don't want to spend any time
- 20:40on that. I just want that product there.
- 20:40Tom Ollerton: So from a work context. So I have a bunch of
- 20:40Anastasia Leng: See, you would want that kind of experience.
- 20:42goals that I want to hit as a, as a, essentially a B2B, sales
- 20:46guy, just what's, what's, what's my role. And there's a tonne of
- 20:50things that is so unbelievably boring that it can just take off
- 20:54my hands. So it makes it does jobs for me that would take
- 20:57hours. So if LLM search and predictive choices and decisions
- 21:01like, the one thing I want, and I can't believe it doesn't
- 21:05exist, is a practical application for this, for email,
- 21:08right? Google has whatever Gemini, but, yeah, I open email
- 21:12and it's like, they've got eight emails. It hasn't even got to
- 21:15the stage where it goes, actually, this one from
- 21:18Anastasia about meeting with Thorsten. This, you need to
- 21:22action this now, am I right or wrong? You can sum it up down
- 21:25yet the the email is based on recency, and we confuse recency
- 21:29with importance, right?
- 21:31Anastasia Leng: Yeah. And no, I agree. I mean, it's everything
- 21:32from email to calendar. We work on a on a sort of First In,
- 21:37First Out model, which is not reflective of our focus and
- 21:42prioritisation at all. And I do think we're seeing some startups
- 21:45pop up. I know and I can't remember the name of one, but
- 21:49there was one which is exactly this, where there is essentially
- 21:52a a shop front to your email, where you can go in the back
- 21:56room and find all the emails, but it actually says to you,
- 21:59these are the the ones you actually treat and everything
- 22:01else, if you want to go to the back room and check them out,
- 22:04it's fine, but these are the ones you need to read. As an
- 22:06inbox zero person, I have to say, my initial reactions a
- 22:09little little nervous, but, but I would, I would try it out.
- 22:12Thorsten Schapmann: You have another question. Otherwise,
- 22:13I'm, I don't. I'm just, I'm just no reflecting on these, on the
- 22:19prompting behaviour. And I like your example, and I think the
- 22:23way, how I assume many consumers will react with LLMs is much
- 22:28more complex. I mean, I get, personally, much more joy in the
- 22:34results when it's much more complex prompting, like you
- 22:38mentioned earlier, from, let's say, well being, and what would
- 22:40be the right recipe and what would you recommend. But then
- 22:44it's again, then how does brand play a role in it? The
- 22:51affiliates, I absolutely think is, again, a very good quick
- 22:56win. So like planning your vacations is something that I
- 23:00like now to do on GPT. So give me the best restaurant. So wait
- 23:06to the best museums around Cannes, where I get tickets, and
- 23:10then, of course, I provide the links and stuff like that. So
- 23:13there is already, I think, a huge use cases where they where
- 23:17is opportunity to drive revenue, but the complexity of problems
- 23:21would probably be the thing that I found quite interesting, and I
- 23:26need to get a head around how brands can play a role in that,
- 23:30versus Google, where it's like much quicker sometimes, and of
- 23:34course, yes, over time, they became questions, and not only
- 23:36words, but now with the LLMs, is really like more complex,
- 23:41conversational discussions, which I personally like as a
- 23:46user interface experience. Yeah. So the role of the brand is
- 23:51something that probably would totally differ from CPGs to
- 23:55travel to insurances, financial products, etc, etc. So that's
- 24:00super exciting to see what's happening.
- 24:01Tom Ollerton: A vision I was talking about with a guy called
- 24:06Michael Cohen just now is that I would like to get to a stage
- 24:10where the the LLMs can organise all the stuff that doesn't
- 24:15require any physical action from me. Right? So to your point
- 24:19about that email organiser, imagine that it was trustworthy.
- 24:22And it works because, like, there's some meetings that an
- 24:26LLM can't have on your behalf, right? You need to show up,
- 24:29like, we're going to go for lunch, breakfast, coffee,
- 24:32dinner, whatever it is that needs to happen in person. There
- 24:35isn't. There is no like, an LLM couldn't do that. So I would
- 24:38love to get to a stage where it's Monday morning, and you go,
- 24:41right, what do I need to do? And it's like, you need to spend
- 24:44time with Thorsten. You need to go to the office in Portugal, or
- 24:47whatever it is.
- 24:47Anastasia Leng: Well, remember, you had this idea. I don't know
- 24:47if I can talk about this, but you had this idea a while ago,
- 24:53of...
- 24:54Tom Ollerton: I'm really worried about what you're going to say.
- 24:57Anastasia Leng: Don't worry. It's a it's suitable for a
- 24:59podcast. So you had this idea a while ago, which was you would
- 25:04have essentially an agent who would screen people who wanted
- 25:08to partner with you or who wanted to sell to you, and then
- 25:11essentially decide on your behalf if you should take that
- 25:15meeting and you like that would be really useful, because then
- 25:19you you basically, you know, you talk to the to the to the agent,
- 25:24to the LLM, it knows what the criteria is, and and all of
- 25:28that. And I thought, and maybe this is because I'm a control
- 25:32freak. I wouldn't want that at all, right, because I still
- 25:36believe that, perhaps naively, that there is, there is a
- 25:39judgement call for some of these decisions, like, Who do I hire?
- 25:44You know, who do I partner with? Whose products do I buy? That is
- 25:48a slightly irrational, this not a perfectly rational decision,
- 25:52right? And so I wouldn't trust an LLM to replace some of these
- 25:56things, no matter how structured it was. Now maybe look like,
- 26:00maybe the proof's in the pudding to your point, if they can do
- 26:05this accurate and get to stay in line with your point of view,
- 26:09yes, but then you get into this much bigger conversation, which
- 26:13is, what are the decisions that are merit to your time and
- 26:17attention, versus what are the ones that effectively do not,
- 26:21and you can cross itoff the list.
- 26:29Tom Ollerton: I would argue that that is already happening with
- 26:32websites. Right? Every time someone goes to creativex.com
- 26:37they're having, they're having they're having a meeting with
- 26:41you, right? That's an automated meeting, right? It wouldn't like
- 26:45every word that's written on your website came out of your
- 26:47mouth or your fingertips at some point to represent you perfectly
- 26:50in written form. So every time someone comes to your website,
- 26:53effectively, they're having an automated meeting with you, and
- 26:57they're making a judgement call whether they book a demo or
- 26:59whatever it is. So I think that is already happening, and thank
- 27:01you for reminding me of that particular idea that I had. But
- 27:04my point with that was Thorsten global VP, so I got my jobs out
- 27:11like you. You can't afford to spend many hours a week meeting
- 27:15new suppliers. You can't because you've got a job to do, but you
- 27:18also can't afford not to, right? So I like the idea of an LLM
- 27:23search, or not, where it's where it is that. So have a synthetic
- 27:26meeting with you, so your face will appear, and I can have a 14
- 27:30hour meeting with you if I want to go, Yeah, but what about
- 27:32this? Do you need this and then, but you've briefed that LLM to
- 27:36say, I, you know, this is important to me. This isn't
- 27:38important to me. And sorry, just before I pass over. So you've
- 27:41really got me thinking. I got a call on the Croisette, and it
- 27:45was like a mobile number from the UK, and this really lovely
- 27:49guy was like, I'd love to introduce you to this product
- 27:52called Cool AI, whatever we're, you know, we're a and he was
- 27:55getting a bit nervous, and you could hear the call centre in
- 27:55the background. He's going, oh, like, but I was like, oh,
- 27:55there's something quite charming about it. I'll give him more
- 28:03than a few whatever. And then, and I was like, Are you? Are you
- 28:09a chatbot? He says, Yes, I am. I'm a chatbot. Then, and I was
- 28:12like, I was like, I work in AI. Worked in AI for 10 years. I run
- 28:17a creative automation business. If anyone should be able to spot
- 28:19it... and I did. But it was like, it was so natural. It was
- 28:23brilliant. And then I booked a meeting with him, so
- 28:26unbelievable. It was quite it was, it was a little magical
- 28:29moment. So like...
- 28:30Thorsten Schapmann: So your idea is already, more or less
- 28:32reality. And what somebody was in Call Centre in London,
- 28:36probably.
- 28:37Tom Ollerton: And I was out, but I was like, Well, what do you
- 28:39do? What's the technology built on? How does it integrate with
- 28:43my own CRM? Like, I was just like, you know, and once I
- 28:45realised there was a robot I could just get, like, there was
- 28:48no, I didn't have to worry about their feelings. And I can't
- 28:52remember the name of the company, but I've got a good
- 28:54understanding of how that works. So that was a synthetic they had
- 28:57a synthetic meeting with me, and I've saved their sales guy, like
- 29:01half an hour an hour, plus all the prep. So...
- 29:04Thorsten Schapmann: I just have another I've just heard it last
- 29:06week on another summit that General Manager or director can
- 29:11not even recall the company, but he was testing to being in on
- 29:17several conferences virtually, and doing a Q&A because they,
- 29:23let's say, programme, the LLM and an agent, and so that he
- 29:27could be representative on, I don't know, five different
- 29:31continents at the same time, doing a Q&A that was threatening
- 29:35to a lot of people in the room, some C level people like, Okay,
- 29:39what does it do with liability? What is it when you say
- 29:43something that is, let's say, legally binding, or has,
- 29:47especially on the general manager C level, has also kind
- 29:51of legal implications. So that's, that was really where
- 29:56you see a lot of goosebumps going out being a wow. That's,
- 29:59that's thrilling on the one hand side looks quite astonishing,
- 30:04because you can do reduce the travel time, be more present to
- 30:09your people. But on the other side is really has a huge trust
- 30:13in what's happening in your name, with your face.
- 30:17Tom Ollerton: Yeah, this is really annoying. We have to wrap
- 30:19up. Anastasia, you're you have something to share.
- 30:22Anastasia Leng: Oh, can I ask? Can I ask a question?
- 30:26Tom Ollerton: I have a flight to get so.
- 30:27Anastasia Leng: Oh, maybe. Okay, so, so you seem not to be too
- 30:32excited about this concept of, like, affiliate links as the LLM
- 30:35model and almost a like, paint the scenario for me. So you log
- 30:40into GPT, and you get what, an ad, or like a text based
- 30:43recommendation based on its knowledge of your feelings.
- 30:46Like, if you had to design an advertising system within an
- 30:49LLM, I get that you'd want it to be predictive and sort of ahead
- 30:54of you based on what it knows about you. But how would you
- 30:55want the ad to actually show up?
- 30:55Tom Ollerton: I think it will be a you know, that you know when
- 30:59you arrive in the kitchen on Monday morning with with your
- 31:01partner, and you're sort of shuffling around each other
- 31:04being smoked salmon or whatever it is. And you're like, What
- 31:07have you got on this week? Oh, well, I've got on, you know,
- 31:09I've gone to seeing Tom to chat with Tom, and I think it will be
- 31:13like that, right? It will be like, okay, chat GPT or whatever
- 31:17your, you know, agent is called. And you'd be like, right? Okay,
- 31:21remind what happened last week, what's going on this week, and
- 31:25then, then it will produce in some form screen, might even be
- 31:28audio about what, what is going to happen, and it will make
- 31:31decisions. And I actually don't think ultimately, there will be
- 31:34an advertising interface. I think actually old school ads
- 31:39served up in a kind of social media format is probably, may
- 31:44well be where it is,
- 31:45Anastasia Leng: Aha. So your prediction is actually that that
- 31:48the concept that the way that the LLMs will monetize is
- 31:52through in is through subscription, and they will be
- 31:56useful because of their embedded recommendations ability to serve
- 31:59you, yeah. And this notion of advertising, the LLM, actually
- 32:02isn't necessary at all.
- 32:03Tom Ollerton: Yeah. And, you know, even now, like, what?
- 32:06What's the pro version of chat GPT cost $200 you know, like,
- 32:10someone will have, like, well, I pay 10 grand a month for mine,
- 32:13you know. And then maybe the advertisers will pay to, you
- 32:16know, you to influence the search or the and I think the
- 32:20affiliate one is, this is the wisest one. People will bid on
- 32:22being affiliate number one, the programmatically. But I think
- 32:26that's the sensible thing. And advertising always just takes
- 32:30the thing you used to do and shoves it into new thing and
- 32:31hopes it works. Anyway. I have, literally, I have to fly. Oh my
- 32:34God, that's been brilliant. Thank you so much. I've really
- 32:37enjoyed this discussion, I think, from three people that
- 32:39didn't know anything about AI search, pretty much 20 minutes
- 32:42ago, we've come up with some good stuff.
- 32:44Thorsten Schapmann: That's the wonderful world of digital
- 32:45marketing, a lot of buzzwords, a lot of opinions, and really
- 32:50looking forward what will happen in the future. Thank you so
- 32:52much.
- 32:56Tom Ollerton: Absolute pleasure.
- 32:58Anastasia Leng: Thanks Tom.
- 32:59Tom Ollerton: hi. Just before you go, I'd really appreciate it
- 33:03if you could take the time to write a review of the shiny new
- 33:06object podcast on Apple podcasts or iTunes, or whatever it's
- 33:10called these days, or whichever podcast provider you use. We're
- 33:14at any podcast. So it would go a long way for us if you could
- 33:17just share the Word and give us a bit of a support on those
- 33:20channels. That'll just be fantastic if you haven't got
- 33:23time. That's also cool. And yeah, if you could tell your
- 33:25colleagues about the podcast, and also, if possible, don't
- 33:29forget to subscribe. And I'd love to hear your feedback. If
- 33:32you'd like to speak on the podcast or be a guest, or you
- 33:35think I'm asking the wrong questions, anything, I'd be
- 33:37super interested to hear what you think. So please email me at
- 33:41Tom at, automated creative.net that's Tom at, I'm not gonna
- 33:47bother spelling it anyway. You'll work it out. Thanks so
- 33:50much.