Ajinomatrix.org CEO Francois Wayenberg: Digitizing the Senses of Taste & Smell

My name is Dustin Betts. I am a community manager at the Founder Institute, and today I'm going to be speaking with a founder at the forefront of digitizing human senses of taste and smell for the food industry. If you're a food or agtech pre-seed stage entrepreneur listening, you can actually apply right now to join the upcoming FI Global Food and Agtech Accelerator at fi.co/food-ecosystems. But without further ado, I'm joined here today by François Weinberg. He's the founder and CEO of Agenometrics, based in Belgium. This is an FI Berlin portfolio company simplifying life for food scientists through sensory intelligence software. François, thanks so much for joining us today.

Hi. Well, I'm like you say, the founder and CEO of Agenometrics, and I'm very happy to be here today to talk about projects.

Perfect. So before we dive into Agenometrics, please tell us a little bit first about yourself or your own startup story. You know, what were the experiences of yours, professional or personal, that led you to founding this company basically at the nexus of food science and artificial intelligence?

Well, originally I'm an economist from the University of Brussels, and I became also in the progress, in the process an industrial collaborator at the AI lab at the Polytechnic Institute there at my university. But after a very amazing project that we've been into, the cyber theater, we had twice a prize from UNESCO, getting the Best of the Web Award. I moved on to a very— unexpectedly, I was contacted by a Japanese trading company I was then unemployed because the preceding project was a victim of, you know, the bubble. So that was quite some time ago, and I was contacted by this really amazing company dating back from 1620 AD, so an earlier company than the East India Trading Company. And there I was European food manager and I had to face an extremely stringent and organized approach to food quality for export in Japan from Europe. And in large part, I had to educate the suppliers in Europe to— for them to be able to export their produce in Japan, to Japan. So actually, it was so much so that it later allowed me to even collaborate with NASA through what I had learned at that company. So my motivations, you were speaking about them when I worked at Mitsui, I explored what was feasible as most advanced with one of the big 5s. And we made up a plan to digitize the senses of taste and smell for the food industry. And I have to say, this really kept me going on for a very long time. And got me thinking because of two things. First of all, I got started while working over there to boot up the project with a former Microsoft speech recognition company from Belgium, also LNH, you know, Lernort en Housepi. And they had an incubator, the Flanders Language Valley. And we started working there. And I met someone who really changed my life, Joel Bellinson, who just had made the COVID of Wired magazine at the time as the inventor of the smell printer, and that was in London, and we discussed his invention and the involvement of what we would like to do And, well, unfortunately, him as well had problems with the bubble, so I continued his research for quite some time over to Switzerland, then to Israel, then to Berlin, and finally came back to Belgium. So what's really nice in this story is that he's like my dad in research, and I'm his son because I continued, you know, everything up to now, and we met quite a long time afterwards, and he said, "What? You did all this?" And we decided to found the company, and he's a founder with me now.

Cool. That's such an interesting story. So you've been all over Europe and doing different things with data as well as with the food sciences industry. So kind of a long road traveled to find yourself here to founding this company ultimately with your the predecessor of the research that you picked up, his grant. So, yeah, super, super interesting. I know because I'm somewhat familiar with it that, you know, Agenometrics is providing easy-to-interpret sensory data so that food science decision-making can be more accurate. But sort of give us your short pitch version for those totally unfamiliar coming to this blank. What is Agenometrics?

So Ageniomatrix digitizes the measurements of the sense of taste and smell for the food industry through an AI B2B open source app, allowing a food company of any size to implement AI and sensory digitization within their facilities, and either for QC or product development. So it solves The problem that nowadays, even in the 21st century, digitization of the sense of smell and taste is a very— it's craftsmanship, it's very artisanal as a process. And even in medium to large companies and even in majors, the digitization level of facilities is not complete or not even generalized.

Yeah, I want to kind of dive into those industry problems and solutions a lot more here. So like, for those who are removed from the food sciences industry, a lot of them are probably totally unaware of the level of precision testing and data analysis that happens in test kitchens and lab benches, all the things that go into what food industry players have to— decision-making around their products before they go to the supermarket shelves. People have probably heard of taste testers. It's one of those jobs that I don't know anybody who really has it, but people imagine one of the best jobs in the world is being a taste tester. So maybe people have a little bit a little bit of inkling, but I get the sense that this is a pretty, like, impenetrable industry space for most average food consumers. So can you kind of explain a little bit more about, yeah, that full scope of food science complexity here and, like, the sort of problems that you're solving for?

Yes, of course. It's funny how you put it because a few days ago I had a a potential investor from Poland, and he talked to me about working in a chocolate factory. And actually, he said it's not like such a dream job because in the end he didn't like the product so much because he had to taste it on the line all the time to say whether it was go or no go. But yeah, how to explain? The players in the industry have to make decisions on how to organize the tests in order to measure the taste and aroma. And they can do it either internally or delegate it to an external lab. Obviously, inviting competent tasters as mouth and noses is not only very troublesome during, for instance, the COVID pandemic, like we are into now, but it's also— it's already so normally, it's very difficult. It's costly, it's not easy to organize, and it takes considerable amounts of time. So how does it work? People as tasters, they give notes. So they are judges. It's like an ice skating performance on a scale from 1 to 10, for instance, and the results are being encoded. And this, in addition to some lab equipment and maybe a few sensors, it's about how the data is currently being recorded. So it's still not a very elaborated process. Our idea on that, it's more like making a sequential file like an MP3 that is able to sequentially represent the experience of tasting because it's like a symphony. It's really like music. You taste a piece of chocolate and it starts to boom with the sugar and the taste of chocolate. Then you have the body of the taste and then it fades away. And then you take another one, and it starts again. So how this is being recorded is still happening in a very old-fashioned way, in such a way that the experiments are not being standardized from one to the next. And as a consequence, the results are not being comparable.

Yeah, yeah, there's still some element of subjectivity, I guess, in that. The only— I guess— This is a very small subsection of this industry, but people are probably most familiar with sommeliers and wine tasting, where it is this kind of thing where, as you described, what's at the front of the palate and then kind of at the end. And that's what happens in the food sciences industry, right? There's professional tasters, and they try to make it quantitative, but there's an element of subjectivity, as I described it. So why a genometrics? Talk to us a little bit more about— so the digitization of this sensory data, why is that like a key solution for addressing some of these challenges?

Well, as you imagine, it's quite complex. I'll try to be short in order not for people to fall asleep. The problem is that because the organization of these tests, how they are being recorded, there is no modern data analysis possible on the results of these tests. And the players in the industry, they have their hands at a large set of data, but that data is just dark data. So what GenoMatrix does is to bring in an easy framework in order to have this data standardized so that across experiments— because it's really, if I can take the simplest example, it's like Dory in the Walt Disney fish story, the clownfish. Well, in the world of Nemo, the experiments from one to another, they are being simply forgotten. And this amnesia is just terrible. So we standardize those data and we help implement data mining and AI in order to make sense of the accumulated data. Because there is so much in there we can spare batches that will be normally thrown away or save time in product development in order not only to interpret it in a classical data mining way, but to overcome the limitations of the classical sensory science. You know, it's nonlinear, it's in terms of the how we interpret the data.

Yeah, yeah, that makes a lot of sense. Yeah, it's definitely complex as you're describing it, but definitely that digitization, I like how you describe between the different tests that concept of amnesia, like there's just without the standardization, it's sort of just lost between. And so yeah, you're able to provide some more continuity through that standardization. So yeah, cool. I want to talk a little bit more, I guess, about the features. I mean, I understand like your software can already interface with some of these like emerging e-notes and emails. I've only heard like a little bit about that new kind of sensory technology, but it sounds very interesting and really promisingly aligned with like your own technologies. So, I mean, are there, I mean, other key features of the genometrics that we haven't talked about or that you want to highlight here in context with any of those like emerging, you know, e-sensors?

I, maybe I'll surprise you because I'll take an angle of approach that is a little particular, because it's a— in fact, it's a delicate matter. It's a delicate issue because we have, for instance, a giant in flavor and fragrance coming to us asking openly how we can use our technology to help them basically get rid of all of these panels with professional testers. Their idea would be simply to start working without those. So I would say we develop the tools and we offer them to our clients. When it comes to interfacing the sensors, also we are not quite there yet in the industry in general. It's projected in phase 2 for the deployment and development, but it's not really what we are focusing on right now because The reality in the industry is that these technologies are in their birth phases for industrial applications, and they don't really replace the human senses just yet. We're getting there, but so there's indeed an alignment and a promise, but we will still move on to these types of proposals when there is a set of agreements that we will conclude with for instance, sensor makers in order to implement interfacing, as at the moment we are dependent on data exports for software to include them. Well, this said, we are in discussion for a large joint venture with an output device for smell, which is something our team has had previous experience with, namely with Joel.

Yeah, yeah, it's definitely at the very, very beginning of some of this sensor data because my understanding is like that there's people talk about the different senses that you can taste, right? Sweet, salty, umami, and stuff like that. And it seems like even with that, like within the last decade or so, there's some interpretation around, well, do we know exactly all of these different parts of the tongue or do we have all the senses even? Like, do we even know what they are? So yeah, early days for that connecting to the e-mouse and e-noses. Yeah, absolutely.

If I can add something, it's that we are integrating the senses. So we do the complete phasing of smelling, smelling in the nose space, in the mouth, and of course, tasting. And the tasting experience is 70 to 80% smelling. So we do all of it. And it's clear also that in the future— and this is repeatedly asked to us— to integrate elements like texture. And also, it can go pretty far because we can imagine to integrate vision because it matters. Of course, texture because if you eat the chips, if it's not cracking, you won't eat the second one. Because there are also experiments going on with the the judges, that is pretty interesting because when you see their reactions, for instance, that can be measured also. The emotional component is very important and we can integrate that because what we do is to make a data repository to characterize food items. So that is something we can look into, of course.

Yeah, that's really interesting because I knew that, like, how connected your sense of smell and taste are. But yeah, I hadn't even thought about texture. Obviously, that's a huge deal. And yeah, like the other data you could get, like the emotional reactions to build like that comprehensive picture. That's really, really interesting. So if I'm a food scientist hearing this right now for the first time, how can I get started using Genometrics? It is launched and out there in the world, right?

Yeah, I would say if you, if you are a food scientist hearing this and are interested, we are actively looking for use cases, including involving sensors. So I really invite you to contact us through our website. We have a calendar in place, you can directly book a demo, which I will probably hold and we will showcase the technology. So that's the best channel we have in place for the moment. And about implementing Genomatrix in your company, that is, it's quite straightforward. We are making a data audit and we work on data lakes and data warehousing in order to realize couples of data between production data, P-Chem, you know, physical chemical, and we bind that with the sensory data. So we have an integrated approach to quality control because, you know, quality is determined in T+1 and quality is determined by production, which is taken in T. So there is a desynchronization. That we are getting through with AI and then we are getting expected behaviors instead of having to wait for the experiments to take place. It's really exciting, so please come.

Yeah, perfect. So if that's you and you're listening, yeah, go over. We'll have the websites linked here everywhere that we're sharing this conversation. And yeah, you can book a demo and see it. Jump on board. This is my next question, is one where I ask you, my guest, to make a prediction, which is basically, you know, be a future prognosticator here. How do you think these data technologies are ultimately going to change food science over the next decade, like 10 years from now?

So I will take again a relatively classic take on this, but we are clearly at the start of a revolution of how data is being integrated into food production. This can be, I insist, a very positive change because there is a real promise of how customization and even transparency, which is taboo in the food industry, and a booming of the different niches that we can see appear of products being offered to the world populations with as many diets and dietary customs that there are, as group of people, people suffering from food-related diseases, people with special diets, different religious habits, all are sources of so many different niches the food producers have to learn to address. So just, I will take an example, a practical one. As of today, how Burger King, for instance, offers a wonderful choice of vegan-based burgers And I was amazed to taste one of those because you can feel there is more engineering of how it tastes compared to a flagship regular meat burger. The taste palette in there is wonderful. It's very vegetable-based, and the plant-based burger homogenizes in where— with a very tasty palette. That's very interesting because there are places in the world where these developments are very welcome. But I will take, for instance, LATAM, you know, so South America. They have a sense, for instance, of how tasty the meat has to be. And in order— and it's their main criteria to choose what they will eat. So as a consequence, if you want to sell a veggie burger over there, It has to perform better than a regular meat burger. I think this is an excellent take on how it's going to deploy all of these technologies. And I think it's just the beginning. And we see how much more of those refined tastes are appearing. It is also clear in my experience of working with Japan and Asian taste palettes that our traditional food is booming with a creativity that is unmatched compared to the past. So to answer the question, it is clear that technology is helping to reduce, for instance, fat contents, sugar, and salt for a determined taste envelope today, will, with the help of AI, achieve unpredictable and great developments in terms of the variety and customization of the food we eat. And also produce greater adaptation of the consumer and the requests from the customers as the choice that the industry offers expands. So this trend is likely to produce great changes in the future. Data will serve these developments because in the end it's the consumer having fixed range physically, of how he tastes and smells food. And for that reason, the offer of new food will have to be constrained by that in some way. So when there is a constraint, creativity has a space to develop, and therefore data will be instrumental in fine-tuning and better satisfying the customers in terms of how they are served. And I think it's very exciting what's coming up.

Yeah, me too. It seems like it's probably the most exciting time maybe ever for food science, really, because of all the emerging technologies that are coming out, like you were mentioning. I mean, it's interesting too how you tied back some of these cultural elements that maybe constrain our preferences around taste. But it's possible there could be totally new flavor profiles that are sort of like, you know, whereas like Impossible or Beyond Burgers are obviously meant to like simulate, you know, the more normal, I guess, like meat products. But yeah, just the food science technologies of the future that might unleash just totally different flavors or things that we've, you know, not experienced before. It seems like a really exciting time for the industry. Yeah, I love the predictions. They're in depth. What advice would you offer to kind of pivot hard back to entrepreneurship here for a very early stage or maybe even just an aspiring entrepreneur who's just getting started? Any knowledge that you've accumulated since working on a genomics that you wish you kind of knew at the very beginning of your founder journey?

Yeah, being busy with this project for so long, it's over 20 years now. First of all, never ever give up on your idea once it gets some level of validation. Of course, you have to go through that, And sometimes slightly pivot, but in general, just don't be— you suddenly, you have, I think, to consider— I don't want to go into ad mode, but I think tools like the Founder Institute help you reach your goals in entrepreneurship because it is such a prone environment to manage all the early challenges you are confronted with. While creating your company that by the time you graduate, you basically do everything by instinct and you get used to make your way towards success. And third point, finally, if I— I think it gets to a very particular environment nowadays when it becomes more and more natural to network with people and achieve most of your desires in entrepreneurship through dedication. I then encourage anyone wanting to make his or her idea real to consider to become an entrepreneur and to found a business in order to make it real. I think really we are in special times and let's hope it really stays that way, but I think we are living amazing times.

Yeah, thanks for saying that again. I like that kind of element of perseverance and yeah, like entrepreneurship is kind of the way to actualize your sort of vision out in the world. So yeah, obviously we agree with that at the Foundry Institute. So, well, final question I have for you is just to wrap up on, you know, where is Unimetrics headed next? I understand you might, you're currently fundraising. If you want to elaborate about that or any other news or final updates to share with the Foundry Institute community or ask for the audience who might be watching here today.

Yeah, absolutely. So we are today, after less than a year of the foundation, of the company in Belgium. We are already operating from 8 different locations around the world on a daily basis. And we maybe, you know, we don't have sales just yet, but we are really on the verge of it. And we are launching our first €2 million straightaway round. That's very ambitious. And it corresponds to our ambition to propose things like a sensory file standard, Like, it's like a Photoshop file of the sensory kind that is digitally modifiable but implementable as a recipe in reality, eventually even with a cost estimate. That is one of the things we are working on in Agilent Matrix Research. Also, we— one very important thing for the future is that we operate in a hybrid mode between a startup and a research foundation. In OSS, you would compare that to, you know, Canonical with Ubuntu. It's something that makes us special in the way we protect the consumer. At Adgenomatrix, we have an ethical manager in the organization, and we are a.org. We want our clients to visit us to achieve better results in offering healthy products to the customers, the consumers. In this instance, I would bounce on what I saw recently about the interviews. Possibly a genomatrix in the future could interface with apps such as bettermeal.ai in order to offer sensory preference and more adapted and healthy diets to the consumer, but based on the, you know, starting with their sensory preferences. That would be very interesting. I think there's an amazing set of possibilities that are just being defined now on how taste and smell digitization and digitization of the experience of eating will interact with us even up to the metaverse.

Yeah, yeah, definitely. Once you add this element of digitization into the food sciences industry, there are ways that you could collaborate with other— all kinds of other technologies. The sky's kind of the limit. Yeah, we're definitely at the very beginnings here. This has been really interesting for me personally. I find this stuff fascinating, learning about the intersection of artificial intelligence and sensory experiences here. And we're all at the Foundry Institute definitely looking forward to keeping up with the Genometrics and where you're headed next. But François Weinberg, thanks so much again for joining us today.

Welcome. And if there is any people from the academic world, we are also in research, so in the United States or anywhere in the world, we're open for collaborations.

Yeah, so please reach out. Awesome. All right, thank you, François. Thank you. Bye-bye.