Beyond Research: How Agentic AI is Rewriting the Rules of Commercial Intelligence

Hello, everyone, and welcome to today’s webinar, but beyond research, how agentic AI is rewriting the rules of commercial intelligence. My name is Caroline Woods, and I’m the marketing campaigns director of Market Logic and your host for today’s session. So for those of you who have been at past webinars, first of all, welcome back. You probably recognize me and, you know, to just sit tight as I go through the housekeeping up top. And for those of you who are joining us live, for the first time, also welcome. This is a pretty straightforward webinar platform. I would just direct you to the q and a tab on the right side of your screen. We’ll be inviting questions throughout, but only hosting the the q&a at the end. So if there is a question for a particular speaker that we have, just maybe put their name in the question so I can better direct them to the person you want responding. Additionally, we will be sharing a recording of this webinar and the slide deck tomorrow, directly to your inbox. So we hope you can stick around for the the full event, but if you need to live early, there’s no worries. So on to today’s event and the topic at hand. We’ve had a lot of webinars and discussions around how enterprise brands are utilizing AI for their market research and insights. But today, we want to come at it from a slightly different angle and look at how this technology is reshaping the work that market research agencies, consultancies, and firms are doing on behalf of those clients. And I think today in the audience, have a mixture of people from the brands themselves, other agencies, and tech companies. So, it’s good to have that mix. I think there’s something valuable in this conversation for everyone. And, of course, us here from Market Logic, we fall into the technology provider bucket. You won’t be seeing a product demo from us today, but you will hear mention of our product DeepSights, which is a suite of agentic and generative AI tools for, built for market researchers. Now on to our speakers. So I’m very happy to have with us representing the market research camp, Simon Nagel, who is the senior client director at Melli. Simon joins us with twenty five years of experience in the FMCG industry with a particular expertise in the pet food category. His background spans everything from commercial strategy to logistics to brand development. And today, I think we’ll have a chance to hear him touch on all three of those pillars, obviously, through the lens of AI. And then also we have my colleague from Market Logic, Natalya Dyubanova. She is our partnerships manager and is the key facilitator to our relationship with organizations like Melli. So before we get into my conversation with Simon, rather than me try to summarize, what is what these partnerships do, I’d rather hand it off to Natalya since she is the expert. And she will explain how we came to work with Melli in the first place. So Natalya, you can take it away. Lovely, thank you. Hi everyone. My name is Natalya. Sometimes I call myself Partner Nat, and I lead our partnerships here at Market Logic. And I spend a lot of my time working closely with strategic market research agencies, technology partners, and consulting firms across Europe and North America. A big part of my role is partnering with agencies who are thinking seriously about how AI is changing the future of insights, not just from a technology perspective, but really from a delivery quality and client value perspective. And that’s really why today’s conversation feels really timely. And I’m excited to hear more from Melli on all the work and all the conversations that we’ve been having for, I think eight ish months. But before that, I just wanted to share, I went to Succeed a few weeks ago, which is a global market research conference here in Germany. And I came away with three clear takeaways that I thought would be quite interesting to share. First one is synthetic research is no longer a future topic. It really is happening now. Seventy one percent of global market researchers will use synthetic response within the next three years. And eighty seven percent of researchers already using AI audiences are reporting high satisfaction in messaging, naming and packaging results. So the shift is real and it is fast. But on that, the second key takeaway for me is that many leaders are still navigating very real questions around accuracy, governance, security, and the role of human expertise in all of this. So the discussion is not just about using AI, but it really is about how do we use it responsibly with the right guardrails in place. The third point that I took away is that I really saw three distinct types of agencies that are emerging today. There’s the tech builders. These agencies have decided that hiring developers, building proprietary tools, and they’re going to be pivoting towards becoming a technology company. It’s very great. Then there’s the waiters. These are the companies that are starting to use AI for emails, for summaries. They’re still assessing what to do with it, but they’re really not yet moving. And then there’s the partners, the smart partners. They’ve potentially tried to build something themselves internally, but they’ve recognized that keeping pace with AI’s rate of change would actually pull them away from the thing that makes them irreplaceable, which is their deep domain expertise. And this is where the third group is exactly where Partner Nat and the team here at Market Logic focuses our energy. We really believe that through DeepSights, we help organizations work more intelligently with the research and the knowledge. But what excites us even more is what happens with those capabilities are paired with real domain expertise, real strong human judgment. And that really is where the magic happens. Because the most valuable partnerships aren’t just about technology. Technology can only go so far. They’re about the technology and expertise coming together to create something more meaningful, more credible, and ultimately more useful for all of our customers. And that’s exactly why I’m excited to have this conversation today with Simon and Melli, and have been part of this conversation for the last few months. Melli brings a really distinctive model to the market. They combine commercial intelligence, category expertise, and creative activation to help big FMCG businesses move from strategy to shelf. In my view, working with Melli since mid summer last year, Melli has taken a very thoughtful approach to integrating AI into their workflows. They’re not just trying to move faster and getting all the pretty toys out there, but they’re really doing it in a way that stays grounded, controlled and genuinely valuable for their customers. And so with that, I will pass over the virtual mic to Callie and Simon to hear how they are taking their AI approach. Yeah. Thanks, Natalya. That was a great intro to, you know, how these these partnerships really work. But, I think now it’s time to dive into the real meat of this discussion. And we have our expert from the market research side of things here. Simon, welcome to the stage. So I’ve given a little bit of your professional background, and then Natalya has given a bit of an introduction to our partnership with Melli. But maybe you could put in your words in your own words to start off, who is Melli and what do you guys do? Yeah. Of course. Thank you, Caroline. Good to meet everybody. Although Natalya did try and steal my thunder there with so to introduce introduce Melli, we get called many different things, which I think says a lot. But, fundamentally, we’re a strategic commercial partner to some very well known international CPG brands. We do we do call ourselves a strategy to shelf agency, which I think indicates the the breadth of the the projects we do get involved in. And we sit somewhere near the intersection of sort of cutting edge category insight, premium design, and as the topic of today, AI augmented workflow. And we work we do operate across most major FMCG categories, but notably, we do have very deep roots, in the pet industry. We’re small, but I like to thank somewhat beautiful senior team based here in the UK, primary clients, such as Mars Incorporated, Britvic. Relie relationships, I know the not only, long standing, but but span global category strategy and market intelligence, right through to retail activation. So if if you asked what I believe makes us different, it’s it’s our operating model or rather our evolving operating model. So we we combine genuine category expertise, unbelievably around hundred and seventy five years of it across category commercial, buying and marketing roles, with a now sort of ever evolving AI toolkit. But as Natalya sort of referenced, we do you know, we don’t just play with AI tools. We’re actually evolving our entire workflow around them. But importantly, with with proper governance and proper brand safety protocols and something very close to my heart, human oversight as as a mandatory factor. So one way we we like to describe this is crafted or augmented intelligence, so a step beyond, artificial intelligence. And, therefore, every output is shaped by human expertise, and powered by AI. It’s not one or the other. It is both working together, occasionally harmoniously, and that’s that’s really mellowing. Yeah. That’s a great intro, and I think, let’s dive a little bit deeper into the topic at hand, which is AI. Because, obviously, you guys are veterans in the industry and are already working with a lot of premier brands and clients. So I wonder if you could put in your own words why you think it’s important to start to have started incorporating AI into your research processes. It’s it’s fundamentally the the world is just changing, and changing fast. A year ago, most teams would still rely on flip charts and post it notes in category strategy meetings. Yet today, a graduate with a chat GBT subscription can produce a category narrative in minutes that would have taken an agency days. That that’s the new reality. That’s the moving baseline. You know, generic large language models are raising the ceiling of what clients can do for themselves. There is there is no point, putting our heads in the sand, in regards to that. And then therefore, they’re effectively bypassing traditional agency need entirely. And sadly, as an an agency, if your output is only as good as what your client can get from a standard chat interface, you’re already obsolete, which I’ve actually I think I’ve got my first slide. It’s and this is what I refer to, almost on a daily basis as the, you know, the just good enough reality. This reality that’s that’s not theoretical. Major corporations are already internalizing activation design and content creation, using AI, and that will only, continue, if not accelerate. You know, the the agencies that historically occupied and monetized, whether it’s data mining, creative iteration, asset production, that territory is being reclaimed by clients. And, you know, a good a good analogy of the shift would be coffee, going back a little bit. But, there was a there was a time when, you know, a proper cup of coffee required time, effort, expertise. And then someone created instant coffee, and it meant you could have something that is just good enough in a fraction of the time if you if you’re rushing to your next meeting. Another analogy would be the microwave in the Michelin star kitchen gives you access to speed, to to volume, but what about the quality? So coming back to your your question, for for me, it’s not whether to incorporate AI. It’s whether you’re far ahead enough of the good enough curve so the clients still need you in the first place. And it’s a bit dramatic, but but that’s what we’re calling the survival zone, and it demands a sophistication that generic tools just simply can’t provide at this time. Yeah. Yeah. And I think you’ll explain a little bit more about the the trade offs between generic and purpose built tools, but I maybe you can give us an overview of your current AI toolkit at Melli. There’s a preponderance of different AI tools on the market, so you don’t have to list all of them. But, basically, what what are the tools you’re currently working with, and why have you chosen to also, incorporate DeepSights into that suite? Of course. I mean, none of this will be a surprise if anyone has been on LinkedIn in the past three years. There’s some obvious choices. We we use Claude and ChatGPT as what I describe as our daily drivers. So that’s the basic workflow. We use them to synthesize strategy, develop copy, and and just basic documentation. So it’s the housekeeping side of things. We also use CRISP and Granola to manage our meeting intelligence, keeping transcripts, and turning client calls into sort of structured searchable reports, whereas before they were scribbled notes on on a notepad. But I think that the the critical point is that these tools are very much general purpose tools. They are they’re powerful. No denying it. They’re getting more powerfully every day, but they are general purpose by design. And and when at Melli, when we need enterprise scale proprietary data to be examined, for example, very large PDFs, large slide decks, in particular, regulated veterinary science information or or multimarket consumer research repositories. The public APIs hit limits, they hit them very quickly. They used to stop after about a hundred slides, you didn’t even know they’d stopped looking. They just start hallucinating because they’d stopped synthesizing what they’d they’d found up to that point, because they basically run out of grounded data, and that and that’s a risk. It is it is a big risk. Of course, that’s where DeepSights comes in. So I think my, my next slide. Market Logic Platform, it gives us something really quite important, and that is this idea of a sandboxed environment, which is very simply, it’s an environment where the the data used to furnish the responses from the large language model can only come from a verified cited source. Basically, whatever you put in the box, nothing else can interfere with the output of that box. And this is a really, really important thing for our clients who want one version of the truth. So for example, if you were to run a general query using a large language model, you might get three different versions of the same global statistic associated with the category insight that you’re trying to present back to the client. If you only present your, you know, your verified cited sources within the sandbox, you can guarantee that you’re you can confidently present one version of the truth back to the client. And so that’s that’s obviously, that’s why we chose DeepSights because it fills the gap between generalist AI and what it can do and enterprise grade regulated, what regulated industries demand. Think we call it the difference between probably right and provably right. Yeah, exactly. That ever elusive single source of truth is something that I think is now becoming something that is sourceable and verifiable. And when people talk about trustworthiness and and transparency in LLMs, I think you hit the nail on the head there. So I think that’s great. I mean, that one question, I think, could go on for another thirty minutes. But, what if we now just pivot into sort of what it looks like on a daily workflow to to incorporate, this AI and and also how it combines with your category, and expert led, category, expertise, so to say. So we have our DeepSights tool, which is a generative AI tool. How would you say, it looks like when, expert led market research is combined with that generative AI? I think we have to look look backwards to look forward. So the the old model was was fairly simple. Agency would would receive a brief. Weeks would pass by as desk research was undertaken and various sort of category diagnoses to produce, you know, a rough scrap scrapbook of information to start the conversation that could be produced, and then we’d argue about the findings. Ai just fundamentally compresses that timeline from what would be weeks into hours and, in some case, minutes. But, of course, the compression of effort without the the expert oversight is just faster noise, and that that is not something that keeps you in the survival zone. So what what looks different now, expert led AI research, you know, a senior professional with twenty plus years of category knowledge, you know, they’re they’re running the show. They are the the pilot, the the architect. They know which questions to ask. They know when the AI output is drifting, when it’s going off brand, and, occasionally, of course, hallucinating. They know before they can even articulate why that a data point doesn’t feel right. And that’s that’s what we’re calling, or what we what is called tacit intelligence. And that is an incredibly high value human filter and absolutely critical as part of this conversation, in in in relation to the combination of AI and, and category expertise. I’ll I’ll give you an example. In some, some recent client work, we we built a fairly substantial seven hundred slide proprietary repository of of pet industry insights. Of course, giving that to a standard API based large language model, it would just completely fail. It would choke after five minutes of effort. And frighteningly, you wouldn’t always know at what point it’s choked before it created an answer. Moving that same behemoth into the DeepSights environment meant every insight was grounded, also cited and retrievable on on a sort of slide by slide basis, which is super powerful from a client perspective. Of course, it was the human expertise that determined what went into that sandbox environment in the first place, how that those insights were structured and and curated, and what conclusions were therefore the right ones for the client to use, from as an output. So the so the expert led AI research means the human’s not just a reviewer of the output at the end. You know, they’re the architect at the beginning in terms of the curated input. They’re the governor and pilot throughout the process. And, you know, AI can handle the scale. Expertise handles the one version of the truth. I I think that we put a lot of pride on, as technology providers on creating tools that are easy to use, but I think, we would be remiss not to talk about the risks that are associated with not having that expertise when you, go and try to, do these projects like the one you just explained. So, maybe you can speak a little bit about the risks for not bringing that expertise to these situations. Of course. And I am hopefully, there’s no one called Toby on the call, but I I call this I call this the Toby with Claude problem. So Toby is our Bright graduate. He you know, he’s got a chat GPU subscription. He’s been asked to produce a simple market summary for a regulated nutrition category, so it’s important. He types a prompt. He gets something back. It looks plausible. He puts it in a deck, and it’s published. So the problem is obvious. He he doesn’t know what he doesn’t know. He can’t spot that the AI has potentially conflated two unrelated clinical studies. He can’t tell the consumer insight is US biased when the brief is for the UK market. He potentially can’t identify that the competitive landscape is is effectively months out of date because the AI training had a cutoff that isn’t published. And in this in this example, we’re talking about veterinary science, and therefore, that hallucination, that risk isn’t just embarrassing. It’s potentially a reputational risk. So it’s it’s super, super important. So the, you know, the risk of not using expertise isn’t that the output is bad. It’s that you can’t tell it’s bad. That’s the real danger. So human governance is therefore not optional. It’s mandatory. Yeah. I think that, there has been some hesitation in different industries about AI being this replacement, but I think the way you just explained it, definitely negates that. And and this human intelligence or human oversight, will continue to be super essential even as these technologies continue to advance. So what we just were talking about is, sort of the the baseline generative AI power, and it is very, very powerful. But now as new features and new tools are coming to market, Natalya’s mentioned up top this idea of synthetic, respondents and synthetic data, which have really taken market, the market research industry by storm and are becoming more and more, standardized. So I wanted to kinda turn and pivot and ask those same questions about, these synthetic data tools. Internally, we call ours DeepSights personas, and anyone in the audience who has, been in the market research field, I’m sure, has seen synthetic respondents, in some variation. So I wanted to ask you how Melli is building personas for these regulated industries using your deep market expertise. Certainly. So, I mean, this is definitely currently more of a playground than the than the DeepSights conversation. We’ve actually and because of that, we’ve actually taken two deliberate approaches, and the comparison is is fascinating, if you like this kind of thing. So internally, we’ve actually built our own vet and pet owner personas using Claude. So six distinct archetypes, each grounded in real world professional behaviors, motivations, clinical decision making patterns, you know, some base data that we had to to build them on. But these were built by a qualified veterinary professional within our team, so not by a a prompt engineer, just just guess best guessing. Because those Claude built personas, powerful as they are, have have an obvious ceiling, and they also have the same risk of hallucination the same way of using any harsh language tool. And that’s where the DeepSights personas just excel. Market Logic’s dynamic personas, are augmented with, external knowledge, but they integrate verified data in real time, in the same sort of sandbox protected, environment. So for regulated industries, this distinction is critical. A static persona gives you an educated guess. A dynamic expert architect built persona gives you a grounded traceable sort of simulation. And so we’re so we’re running a parallel evaluation of the plaud built personas against the DeepSights dynamic personas, which you’re willing to buy off us. We’re testing the response really on the quality of, you know, the technical veterinary queries that we’re putting to them, you know, the whether the the geographic calibration across markets works in in the two environments. And then fundamentally, we’re we’re putting concepts to these personas to to test, yeah, how the responses differ between the the two platforms. But the no surprises. The early result is quite clear. So for regulated technical topics, the sandboxed enriched personas deliver fundamentally better output. So the the the key insight really is that the the persona is only as good as the expertise of the as the person who builds it. So if you can mirror the expertise within your team with AI, that’s that’s the best place to start. You know, we don’t give a a persona design to a generalist. A vet would build the vet persona. A buyer helps build, the buyer persona, and the AI, therefore, amplifies expertise. It it doesn’t replace it. Yeah. It’s it’s the old adjective adjective, like, quality in is data quality out. So if you’re bringing that expertise as you build the personas, you can expect a much higher level of of accuracy and depth. And, of course, you have a lot of experience in a regulated field of veterinary and pet food. I would just say that, like, we also work in, similar verticals like health care that are equally regulated, and that ability to scale scale while building nuanced personas is, you know, for some foremost, our main goal when we launch personas. But now let’s talk about the risks. You know, we can’t we can’t just wax poetic here. Do you find that there’s any risks for, charging forward with these personas with not without a proper foundation in that expertise, and what would that look like to you? Yeah. I mean, certainly. And the the old adage is definitely true. I think my version’s slightly ruder, but the the risks are actually much more significant with personas than with general AI, and there’s actually really logical reason behind that. Because personas almost by definition create a false sense of authority. So when you get a response from a veterinary professional persona, you naturally trust it more than a generic chat GBT answer. It sounds weird, but it’s true. You know, it sounds authoritative. It sounds grounded. But if the persona was built with someone without the right expertise, it’s, I mean, it’s fundamentally just a generic large language model in a lab coat. You know, it’s it’s nothing more. If I share this slide. So if you if you look at the, the risks associated with these personas, they’re probably probably threefold in terms of the sort of the macro categories. So firstly, geographic miscalibration. So it you know, it’s very easy. It’s not easy, but it’s easy to mistrain a persona based on specific market data. Let’s say use US at market data, and it will then your persona is gonna give you answers based on the US even if you’re asking a question about Japan. And that’s a fundamental flaw. It doesn’t tell you that it’s giving you an answer based on the US. Secondly, and again, very, very importantly for for some of our clients is clinical accuracy. So whether this is veterinary or or health care categories, you know, a persona that confidently states something is incorrect as a clinical position is worse than no persona at all in the first place because it’s it’s you’ve given it authority that it that it hasn’t earned. It just didn’t exist, which is which is risky. And finally, it’s this false consensus. You know? As personas are evolving, we’re building them into, like, almost focus group panels. And, of course, as we’ve all interacted with chat DPT, it can create, like, an illusion of validation when in reality, you’ve just asked the same model, the same question from slightly different angles, except you created your own echo chamber. So an ungoverned persona isn’t a research tool. It’s it’s more of a confidence trap. You think you’ve tested a concept, but you haven’t. You’ve just received a plausible sounding echo of your own assumptions, which clearly is not a valuable exercise. Yeah. I think, you know, the introduction of AI into market research has certainly sped up a lot of things and smoothed over a lot of processes that used to be labor intensive or, very take a lot of time and, investment, but that is not without introducing some new issues. And we might have to steal this slide from you after the the deck because it’s it’s a really good summary of of sort of the difference between DeepSights and some of the generic nonmarket research built AI tools. I think let’s maybe, sort of try to conclude this, and then we will move into, like, the audience q and a. So if anyone has questions for Simon and I will also bring back Natalya, please just throw that in the q and a box. But if you could just summarize then, Simon, why do you think or how do you think it benefits clients to work with an agency that combines both specialist AI and deep industry experience? I think probably the the the best way to articulate this is it sort of bring it back to, you know, that just good enough curve and this this sort of fundamental, albeit slightly scary shift. And my boss might shoot me for saying this, but the reality is that the billable hour is collapsing. You know, when when when AI can execute a week’s worth of research in ten seconds, time is no longer a meaningful unit of value. You know, that age old social contract, time equal values, time equals value, it’s broken fundamentally broken. And the agencies that survive will be those that stop selling hours and start selling, outcomes. But, of course, outcomes themselves require, two things that generic AI cannot provide, by itself, and and this is where, you know, the the agent the agency retains its value. So firstly, it’s the secure infrastructure. It’s sandboxed environments where proprietary data stays proprietary, where every output, is cited and traceable, and where enterprise grade governance is is built in. It’s it’s just not bolted on or an afterthought. Secondly, what we talked about earlier was this tacit human intelligence, something it’s hard to really pin down, but we we know is real. That deep sector specific expertise that tells you when an output, an AI output is is brilliant, but more importantly, when it’s dangerously wrong. So it’s it’s how we deliver this value for clients. That’s the real opportunity here. And we have we we’ve actually we’ve put a label on it, because it’s important to us, and we call it the the concierge model. And and in the you know? So in in in this in this new model, the agency effectively bears the enterprise level technological investment. So, you know, we hold the DeepSights license. We we have the sandbox infrastructure, the governance frameworks. All that heavy lifting is done, and we’re able to amortize that cost across multiple client projects, which means brands of all sizes get access to AI capability that would otherwise require, you know, six month internal procurement cycles, probably more in some cases, you know, security audits, dedicated internal teams, the training, and so on and so on. Whereas what we’re able to do as the agency is absorb all that complexity and offer, you know, a plug and play off the shelf solution and access to these powerful tools. So clients who work with agencies built in this way get get, you know, three things that you cannot replicate internally with the chat GPT subscription. So you get the speed without the risk because every output is is governed by, you know, the category experts. You get the depth without the cost because the infrastructure investment is is shared and and not multiplied or duplicated within the organization, which I think I do have one I have one more slide. Was this one. Apologies. Already on it. It and so we’ve we’ve fundamentally, we’ve rewritten the the value equation for, if we’re gonna call ourselves a market research agency at this at this at this point, that the the equation is quite simple. You’ve got your proprietary sandbox data that, you know, is in very safe hands. You combine this with the the tacit human intelligence in those hundred and seventy five years of category leadership knowledge, and you that gives you a decision ready output. What we fundamentally believe is that that’s the future of agency value. So the billable hour is no longer relevant, and that’s why I feel that the Melli and Market Logic are really collaborating on and building together. Yeah. Great. I think with that, let’s bring back Natalya, and we can have a quick wrap up with just the audience q and a. And, yeah, it’s great to have found, like, the synergies between Market Logic and and Melli. I’m wondering, Simon, just a a final question for you. You have this proposition of how agencies can now deliver new value in the in this, you know, new era and new way of working. Do you think that the client expectations have changed, since AI has become more frequently adopted? Do they expect you to be fully in a AI enabled? Does that change the speed at which they expect you to deliver project or even the types of projects that they request from you, or is there still sort of a lag between what’s available and what’s expected? Yeah. The simple answer is we’re in a transition phase or transition period. It does vary enormously. You know, it might be we receive a very, obviously, AI engineered brief from a client, which indicates, you know, you know, a warm environment for an ongoing AI collaboration to deliver their project objectives. But, of course, there are still those clients that are nervous, and they’ve not yet had the opportunity to understand how we’re able to solve for some of the risks associated with the large language models, you know, that they’re not familiar with tools like like DeepSights or DeepSights personas and how we can actually we can achieve, their objectives in compressed time frames, but still offer the same governance and expertise and oversight that we would, with some of the more traditional approaches to their to their projects. And so it’s it’s very much it’s one conversation at a time. What what, of course, we’ve, very much, deliberately and proactively aspired to do is get ahead of the curve so that we can confidently have and confidently have those those conversations. Yeah. I think, Melli has been very proactive in in how you guys have approached AI, and I think you would be a great model for other agencies trying to do the same. But I’m wondering if, Natalya, you could chime in. If there is an agency, that’s looking to be more AI enabled, what do they need to get started? What would you, suggest to them? Yeah, I think similarly to how we started the journey with Melli, and maybe it felt a bit too early and too soon back in June or July last year, but it’s introduction of collaboration of understanding how does your agency work? Where do you think there’s gaps in the processes? Where are your customers challenging you? And I think it’s quite cool being a company where we have the flexibility to be creative. So we’re always listening to what our customers are requesting. We’re always listening to market research agencies or consulting firms, because obviously they’re the subject matter experts at what should be out there. And we’re also watching the industry. So we are able to collaborate on new ways and new approaches to try new initiatives with AI. And so essentially kicking off an introductory experience to understand how do we collaborate, how do we bring synergies, and then going from there. And maybe I pass that over to you as well, Simon, as you’re, very much an AI expert within Melli. But if you were, just getting started today, adopting AI tools, where would you start? What what kind of advice could you give to someone who’s maybe not as mature as Nelly is? It it obviously, it does depend enormously on your, your field of expertise. But the the the one thing that we’ve embraced above anything else is is just the spirit of having an open mind, and that, you know, that that comfort with, you know, not just the not, I’ll call it, enforced change, but there’s that sort of organic and very obvious change because the market needs are shifting. But we, you know, we we talk about, again, it’s slightly dramatic language, but the sell your house moment and that some there are times where you’ve gotta make very proactive and firm decisions to change your operating model. And and, of course, there is an element of we’re we’re make placing a bit of a bet here. Now is the right time to start making different bets because no one can completely predict what’s gonna happen over the next six months, let alone the next three years. One thing we can be certain of, it’s going to look different, and, therefore, it’s the right time to be having a very, very close look at how you deliver value for your clients, where you may need to change how you frame that value for your clients and potentially deliver it with tools like the Market Logic DeepSights suite. Yeah. Great. I think that’s a a perfect note to end things on, especially as we’ve just hit the forty minute mark. So I will say, this conversation of AI and market research, it’s not going anywhere. It’s just gonna continue. I mean, we’ve started to allude to the whole conversation around synthetic personas, and I think maybe we’ll have you back in six months, Simon, really dive deep into that, Simon, because that is where, the industry is is going. But as you’ve heard, it’s a conversation that, I guess, can sometimes feel a little bit existential, but it is also very exciting. And from our side at Market Logic, it’s really rewarding to find the ways that we can work with companies like Melli, and we look forward to building both the relationships, with these companies and the technologies, that underpin them. So I think, lastly, thank you to Natalie and Simon for sharing their perspectives today. Thank you to everyone who joined live. I’m just going to quickly link, to the Melli website and to our partner website. If you want to explore more, we are very easy to find and get in contact with. So here is the Melee website that you can click to go to. And if you were interested in exploring our partner ecosystem, you can also head to Market Logic Software dot com. So, yeah, give a wave, Simon and Natalya. It was a very interesting conversation, and, we will be in touch with the recording and the deck tomorrow morning. So thank you, everyone, and have a good rest of your day. Bye. Thank you.