IIEX Asia Pacific 2026: How Fonterra uses Agentic AI for Insights-Driven Innovation

Yeah, thanks a lot. My name is Olaf Lenzmann. I’m a co founder and I’m running product innovation at Market Logic Software, and I’m very delighted here to meet with Tim today. Hi, I’m in this really weird situation. You guys can hear okay? Yep. In this really weird situation where I work for one of New Zealand’s biggest companies, but have a corporate head office out of New Zealand’s quite a rare and strange thing when you’re in a country of five million people. I work for Fonterra, which exports well, pretty much the world’s largest dairy exporter to the world. Excellent. And then, I think this is the thing. You’re good. Excellent. I just want to say a couple of words about who we are as Market Logic, and then I’m gonna hand it over to Tim for the bulk of the presentation, and I’ll come back with a bit of an outlook. What we do in Market Logic is we provide software for our customers that helps them to bring together, consolidate, centralize all their research, in house research, syndicated research, public information, structured, unstructured, helping them to democratize it, but also now using, and I have to use the AI word, I’m sorry, using AI, for example, to proactively look at all the material coming in, connecting the dots, making sense of what’s new, what’s changing, or bringing it to life with synthetic personas to talk to your virtual customers before you enter into research, etcetera, etcetera. And we have been on a journey since, I think, more than a year now together to think about how can we take the next step from, well, just having all those lovely knowledge and insights assets into bringing them into life for an innovation use case. And, I think we made great progress, and that’s what I want to hand over to Tim for. Thank you, Olaf. We might go back and just do a quick prelude to the story, because we’ve been working a lot with Olaf and team on building this front end innovation studio, and you’ll hear AI referenced many times. Again, apologies. But our journey with, the prelude to the story is our journey with Market Logic started in twenty nineteen. Fundamentally, the brief was really simple to Market Logic, which was how do we better sweat the value of all our insight assets? Faster, smarter, cheaper, and so forth. And so the start of the journey was in twenty nineteen, and, you might have heard, I think it was talking to it in the very first presentation of the day, what did she term it? Building building your treasure trove. That’s fundamentally what the start of our journey was. It was nothing fundamental, nothing revolutionary, but just building the treasure trove of base insights so we could sweat the value of them and, unlock value. So like all good insight professionals and trying to be as commercial as we can be, we obviously set some pretty lofty targets around that. One was to double our insight asset value, which was circa around fifteen million dollars worth of reuse of our inside assets. The second one was reduce research duplication and lower spend for agency roster. The agencies get really scared at this point because they think as though this is a lower role for agencies. It actually isn’t. It’s about optimization of the role for agencies within the network. So we obviously, as you often do in client side business, have to show that you can deliver OpEx savings to the business, so there’s a marginal OpEx saving there. But actually, that pillar there was more around re divergence and reallocation of insights spend into an in house on demand insight generation model within within our own business to drive efficiency, and then strategically freeing up that that agency money for much more strategic added value tasks with agencies rather than getting paying agencies good money to do really basic work, basically. So that was it. Third one was removing human time to answer key business questions. Still a still a big problem, do say, in a number of organizations. And then the third one, the fourth one which excited me the most was reallocation of time. And what that one basically was about was basically taking our insights human capital, the most precious thing we have in our organization with which is our insight teams themselves, and being able to enable them to reallocate their time from being tactical, reactive market researchers who are just taking briefs from teams and answering those briefs, to being proactive, strategic business partners who are fundamentally there to provoke, to challenge, and to shape and create the future. So that’s the one that kind of most excited us. And so how that translated is Insight Farm was then born. Insight Farm, we’re a dairy company. Farm, okay, makes sense. But basically, farming for insights was the platform that basically became our treasure trove, if you like, referencing that first thing. Was it fundamentally revolutionary? No. Did it solve a real and meaning meaningful compelling problem? Yes, it did. Did it did it deliver to those last two KPIs of reallocation of time removal of time amongst our marketing, our innovation, our sales stakeholders? A little bit, but fundamentally, this first tranche of development if you like was a find it problem, solving for a find it. It still didn’t fundamentally solve for giving answers to business questions. So the real game changer happened, I don’t know, when DeepSights launched? Two years? Almost three years ago. Almost three years. There you go. Time moves. Right? DeepSights was launched through Market Logic, which was obviously their multi large language model, GenAI edition overlaid onto our, Insight Farm platform. And this is where we started to see answers to business questions really start to come out. So some pretty, so far as our business is concerned, some pretty significant, uplifts, sixty percent increase in unique unique users accessing the platform across the business. In a four month period, we were starting to get numbers like thirteen hundred questions asked and answered of the platform across our across our business, which given our business has kind of been born born out of a commodity business, we thought that was pretty impressive. Lots of hours saved in manual search again supporting that reallocation of time. But again, the one the one thing that I think we were probably most proud of in this journey was the fact that we were starting to kind of shift and enable our insight teams to move from that tactical reactive approach to insights to being strategic business partner. But there was still a limit, you’ll hear lots of buts in this session, lots of buts and lots of ands. There was still a limit to that level of productivity that could be ascertained because what I would say is that even with the overlay of AI enablement through DeepSights, Our insight platform platform was still set up to fundamentally answer the now. And that kinda made sense, right, because a lot of the content that was feeding into it was content that was based on known and established insights, stuff that was looking at existing answers to existing business questions, stuff that was dimensionalizing problems that could be very clearly articulated in categories that we knew and understood well, and cultures that we had good context around and so forth. But around this time, and I guess this is circling back to Olaf’s point around a year or so ago, the focus of my team changed from insights into foresights and to driving front end innovation. And so we needed to make a step change in terms of how we thought about things. And so basically, we needed to start to move insights, making sense of what we know, making sense of what we don’t know, and getting into foresight generation. And so it’s been a really interesting journey for us because we started this journey basically starting on getting answers to business questions, and now ironically, we’re going back to saying we need to ask questions again. Because obviously, foresights is more about asking questions than it is getting getting direct and overt answers. It’s more around dealing less about critical certainties and being able to navigate, and think about critical uncertainties of change, about thinking about how we can take how we can take, and understand, through drivers of change, future worlds that may exist tomorrow that don’t exist today, and create scenarios around those and think about precarious and preferable futures we might decide to play within those, within those future worlds. So it was a big fundamental step change, in terms of thinking about our ask of Market Logic in terms of being able to support a foresight driven innovation agenda rather than an insight one. So we reached out to the Beach Boys, and asked a few wouldn’t it be nice questions. I don’t actually know the Beach Boys, I’m too young. I’ll just admit that right now. But we asked a few wouldn’t it be nice questions, and this I guess was part of the segue in the conversation with Market Logic. Wouldn’t it be nice if we could proactively leverage AI to scan weak signals of change, emerging drivers of change, and so forth, and understand how those were coexisting with one another? Wouldn’t it be nice if we could see emerging trends as they appear and then to be able to track and monitor their development, and also their interaction with categories, with channels, and so forth? Wouldn’t it be nice if we could leverage Agentic AI to basically start to identify white space opportunities within our existing domain of dairy, but also looking beyond dairy and into adjacent categories where we could create new growth opportunities to create new roles for dairy. And then wouldn’t it be nice to not just identify those white spaces, those white space opportunities, but to be able to then basically be able to jump and refine, explore, co create those white spaces, and to be able to generate an idea and a concept pipeline that could actually solve for those opportunities and fuel our fuel our front end innovation aspirations for the mid to long term. And, of course, what we also needed to think about doing is how could we do this within one user experience that basically means people aren’t having to jump into multiple multiple platforms and so forth. So this is in, this is in very early stage development because literally we just started beta testing this, what, three? Two two months ago? Three months ago? But this gives you a bit of a feel for, what we’re starting to push into. So, again, leveraging Agenetic AI to basically be able to identify, and this beyond dairy first because obviously we can’t think about mid to long term innovation through a dairy lens, Be able to identify new and emerging trend areas and trends that they might manifest into new opportunities. Creating white spaces, so taking healthy aging as an example here, for instance, where we can identify emerging signals that that lead to new white space opportunities and then to free up our inside team’s time to basically be able to go and start to engage internal and external stakeholders to really be able to pressure test those spaces, explore them, and so forth. So a couple of the ones that are really interesting for us at the moment, for instance, sleep quality and, cognitive resilience, multisensory functional dairy snacking for emotional regulation, protein fortified convenient formats for GLP-1. So, again, what’s been really interesting in this is being able to see things like the emergence of GLP-1 drugs and so forth, and then be able to see how our, front end innovation, AgenTik AI agent, is starting to basically be able to identify some of those emerging signals and themes and to be able to push, new adjacent opportunities for us to think about and create new roles for dairy within that. And then obviously to be able to kind of move from, prioritizing opportunity spaces to actually start to, co create and develop ideas. Afternoon clarity micro shots, GLP-1, companion guilt free indulgent gummies. Doesn’t that sound nice? Sleep Sentry Premium Truffles. Again, challenging how we think about, accessing using our core ingredients, new consumption opportunities and moments, being able to adapt dairy into novel novel and new ways. But also being able to do it in a way that basically, if you take this one, basic one of the of the developments we’ve been talking about is basically to be able to leverage the agentic AI system to be able to flag and identify when it’s appropriate timing for us taking some of these ideas and pushing them into the pipeline. Because I’m sure a lot of you will know timing’s everything when it comes to ideas, and we’ve had no shortage of great ideas in our business that have often been blamed as bad ideas but simply because the timing’s been wrong, the cultural context hasn’t been ready, market readiness hasn’t been there, and so forth. So some of the smarts of AI here is making sure that we can leverage AI to be able to monitor conversations and emergence of conversations to be able to reflag when it’s appropriate to take some of these ideas and start to consider bringing them back up into our into our innovation pipeline. So the front end opportunity as it exists today, a lot of it is around hours saving and so forth, because I’m sure like a lot of you, particularly on the client side, you spend a lot of time planning and prepping for ideation sessions and everything, and this is this is cutting a whole lot of that time out. But the thing that interests is probably interesting me the most, and I guess some of the the challenge we’ve been giving to Olaf and team, is how do we do a couple of things in terms of challenging and evolving this? One of those is expanding the content that’s feeding into this to move from existing insight content into better foresight signal content. And that’s interesting for us because a lot of the insight that we’ve had sitting with an insight farm has been your typical consumer customer insight demand generation content. Now what we need to do is really focus on signals beyond demand. So looking at where there are emerging science and technology applications we could lift and shift off for long term growth. Drawing on things like patents data, new start up investment funding data, where’s new start up investment funding going? Some of those some of those data sources that provide better long term longer term signals of change. So that’s one piece where we’re really starting to kinda challenge the data that we bring into bring into this insight studio, in this innovation studio. The second piece, and we’ve spoken a little bit about this morning, is bringing our audiences in a very live and engaging way into every front end innovation workflow step based on what Olef and the team’s been designing. That’s both bringing the real audiences in, but also bringing, synthetic, digital twin audiences, AI personas, and so forth, so that we can engage in a dialogue through the innovation process because how many in this room know innovation is not linear, right? And so you’ve got a front end innovation process, but you jump around all over the place all the time. And the credibility of bringing our audiences into those conversations rather than just relying on static insight and foresight content is going to be a really important part of the journey. So with that in mind, I might hand over to Olef just to talk about some of the developments. Yeah, thanks Tim. So, of course, this is and continues to be a big journey, and as Tim mentioned, we’re just concluding beta testing, want to say, and we’ll bring it live shortly. And I just want to talk a little bit about where we are going more from a perspective of building on that capability as next steps along the lines of those items. And for the first element, Tim has mentioned bringing in more data in terms of roadmap. Of course, what we want to do is we want to broaden the base of what we can actually feed into that system. Right now, it sits atop of insights forum, which brings in all the in house studies, the primary content, syndicated content, etcetera. Brings also content in more as in static reports from certain other trend foresights agencies, but some of them have, for example, also their own AI fight experiences, their own tools which can be interrogated, so making that also part of the entire journey and leveraging the full capabilities of those will be very important to really make use of the best information that’s out there. Also, demand spaces is a concept many of our customers use, or other strategic innovation assets where you have rich research, rich respondent level data across markets, across occasions, etcetera, which can also then be analyzed at a much more structured, at HOT level. So that’s one big area of improvements we’re working on. Another one, also along the lines of better broadening the basis of information, is to bring in the ability to query new product databases. Of course, you want to know what’s going on out there, and that can help a lot to find also supply side white spaces and gaps, that can also help to drive the differentiation opportunities, but also can, here and there, maybe give some inspiration, what seems to work well, and what can be incorporated to your thinking as well. And of course there is dedicated products out there, dedicated offerings that exactly do this, and this is what we’re working on to integrate that so that you as a team working on the innovation can seamlessly within this journey make use of the information and no one have to like step out, go to a different place, do that kind of research and come back and synthesize it in your head again. Right. For the second aspect, how do we optimize and validate what we come up with in this journey? What we’re doing is, what we already do as of today, is we enable our customers to use persona agents, as we call it. So essentially, we allow them to create AI conversational partners that are grounded in their research, actually, to bring to life attitudes, perspectives of prototypical segments, and this is something we’re integrating now also into this innovation journey, so that as you’re going along, as you’re working on ideas, as you’re iterating on them, checking them, testing them with AI, for example, against the knowledge and insights you have, you can also kick off a virtual focus group, if you will, that does a persona interview with the AI and comes back with the findings. Of course, the idea is not to say this is going to replace your consumer research. The idea is to accelerate the iteration and go into the actual research with a better, sharper proposal that’s more on point. And lastly, once you’ve done your optimizations, of course, you still need to do that testing, and also here again, being able to drive the testing right from within this flow so that you can, for example, initiate idea screeners, concept tests right from within the environment, get the feedback into it, can further iterate and optimize, so we’re working, we have also launched collaboration with ZAPI for example, and working with others to bring like the best in class and best of breed offerings in the market for this kind of digital research into that. And the ultimate aim is simply to be able to support the teams on this entire front end of innovation journey, but to do it in a way, and that’s maybe a closing thought from my end, to do it in a way that is not aiming to now replace human creativity and expertise and experience with AI. It’s not like you press a button and then, whoo, here’s five great concepts. All you have to do is to launch them. Not at all. Quite contrary to what many people expect, and also an interesting learning, for example, from the beta test is people hear AI and they think, oh, I press a button, it’s gonna do something, problem solved. But actually, it’s quite the opposite. You don’t have to think less, you have to think so much more and harder because what it comes back with is now the full picture with all angles and lenses and options and dimensions that you probably, without the AI wouldn’t have been able to even prepare in that richness, and now it comes to you with this rich set of options that you really need to think through and judge and then steer and give direction where to go next. So I think this is, aside from the technical journey, also a big journey in terms of learning how to work with AI in this AI human collaboration, due to it in a team, not as an individual tool, but rather in a multi stakeholder environment over a period of time. And really, I think, on all of us learning how in the new world of AI to support these kind of long running processes, reaping the benefits, but also retaining and elevating the human unique capabilities in there. Yeah, and that’s what we’ve been working on, and I think we’ll turn it over to some questions if there are any. Does anyone want to ask a question live, or you can ask a question on the app? Question, question, question. So how did you get AI to generate the four sites? Where are these data sources to generate these? Well, I can maybe speak about the part that is more the AI part of it, and can for sure elaborate on the sources specifically. I mean, we use kind of a multifaceted approach. One is we take first and foremost all the information that’s hooked up to the system and comes in on a continuous basis, and then AI is running, I always like to say always on in reality, of course, it’s like looking at it on a nightly basis, checking in, checking what’s new, and cross correlating that with everything that it has seen already and trying to figure out where is maybe deviations, where is new things emerging, where is stuff we need to start tracking now and building information around that. Of course, we hook up to upstream sources that specialize in, well, I would say finding the raw signals that then serve as the input. So what we don’t do is we don’t, for example, do large scale social monitoring, etcetera. We rather rely on other specialized capabilities that do exactly this and deliver then certain refined signals. And I would say that’s probably the big change at our end is we’ve been focusing a lot on unstructured insight content, and now we’re almost moving to structured content and pulling that in. Some of those other examples, know, patients data, series ABC startup investment data, e com data, social analytics, research and grants data in terms of understanding where investments being put behind long term R and D investments and so forth. Some of those sources bringing that together is really gonna kind of help unlock that mid to long term confidence space. And there was a couple of other questions. I’m to off over there. I’m sure you can answer. Yeah. Can can I start with the one? Thanks for the the presentation. The one thing I want to know is how are doing this at a global level relative to, let’s say, country there. So I’ve got innovation that may affect Thailand. Yeah. It’s a bigger egg ship globally. Yeah. So no. Very, very good question. Segues nicely off that earlier. The, the earlier session, right, in terms of the cloak one. So how we generally look at it is we will have our priority regions for growth, and so generally we will be, training and leveraging AI to kind of go and prompt and overweight emphasis on those regions. And then once, big thing for us is identifying scaled opportunity spaces that present scale opportunity across multiple regions. But then as soon as that next level under, that’s when the localization really comes in. Kind of those screenshots you saw towards the end when you start to go from that opportunity space into an idea expression of what that might look like, That’s when we’re starting to leverage it to kind of tease out where those local kind of differences are and staying open to that. One of the key things we’re trying to find here is where we can find formulation and R&D scale leverage across markets, but often where the expression of that idea will be very different locally in terms of marketing, sales activation, potentially the product format that sits in, that kind of stuff.