COSMOFACTORY

Deanna: [00:00:00] This episode is about precision innovation in both the products sector and the ingredient sector. It's about mobilizing technology to bring transparency to the cosmetic ingredient supply chain. It's about automation, artificial intelligence, and a new era of creativity. On today's episode of the CosmoFactory Podcast, I'm speaking with Karen Ho, co-founder and CEO of Gravel AI.
Karen, welcome to CosmoFactory.[00:01:00]
Karen: Thank you very much for having me.
Deanna: Yeah. No, you're welcome. I'm, I'm glad that you're here to talk about tech with me. Um, there are, of course, some very well-regarded beauty ingredient makers and contract manufacturers that already know your company, but for listeners who have not heard about gravel ai, can you tell us, uh, when you launch this platform and the basics of what it does?
Karen: Totally. Um, we are a, uh, two and a half years old industry intelligence platform for the personal care and beauty industry. We take the perspective of a single ingredient and take it through the entire value chain. We take an ingredient or multiple ingredients and we can identify the, um, the sellers of the ingredients. Uh, we can identify which contract manufacturers are using these ingredients in what way, and we can identify the brands working with these contract manufacturers, and we can identify the [00:02:00] fast growing areas of growth for fastest growing areas. For these chemicals, ingredients, and we can identify all surrounding insights related to these, uh, beauty personal care ingredients,
Deanna: Yeah. Wonderful. And so then thinking about the use cases, who, who is your customer in general terms and, and what are they doing with this platform?
Karen: two types of customers, the sellers of ingredients and the buyers of ingredients. So the sellers, for example, and ingredients supplier, they can. For the first time in the industry, they can link the scientific claims of their ingredients to the success of the actual beauty products. For example, I am a hair loss ingredient blend supplier, and I create ingredient blends that help people, um, um, with um, hair thinning issues, hair thickening. Issues or I'm a then, um, [00:03:00] our engine can say, oh, these are the two, three scientific claims. And then the hair thinning, um, claims come up in 40% of all positive reviews of the products that use this ingredients. So we can match the drivers of the success, uh, the drivers of the positive reviews associated with a product. With the scientific claims of a product, and it's not just for, uh, active ingredients manufacturers, it is also for, uh, like a polymer, uh, manufacturer. So if, uh, a polymer blend. Create a certain texture. Okay? It is very rich. It is very smooth. And then we mine the consumer reviews and identify the percentages of those very positive reviews that are linked to those scientific claims.
And we also identify the contract manufacturers that are working with different brands that use their ingredients so they can go to market really effectively. In the past it was an opaque [00:04:00] industry, so, um, uh, it is impossible. Um, prior to Gar AI to know which brands are working with Manu, which manufacturers, but now, um, you can, um, you can sell effectively into the industry.
Uh, our customers are also brands
and the biochemicals. So they can, um, uh, identify the fastest growing trends in ingredients, uh, which are the top, um, surfactants or. Active ingredients used in haircare in the us um, that has certain marketing claims that it, it is a complete, super granular, um, industry and product, uh, ingredient trends, um, platform for beauty brands for NPD and also marketing.
Deanna: Yeah. Yeah. No, that's fantastic. That was a very well explained, explained, uh. Um, look at at who your customers are, but as you suggested, much of this was opaque right in the past. So clearly some of the information that you are making available really was invisible, uh, before a platform like yours existed.
Can you help us think more about the [00:05:00] positive impacts of this kind of visibil of information?
Karen: Absolutely, actually, um, um, the, the entire industry is becoming more transparent. Um, we are just one of the multiple forces that is making this to happen.
Um, um, first of all, uh, a lot of brands, they, um, are going through this, uh, tariff, um, um. Turmoil. They want to find contract manufacturers that are in different locations.
And then, uh, one of the use cases is that they can use us to find, um, contract manufacturers that are good at specific things, um, that are in the US or in Europe. And the second is, um. It can, uh, uh, it, it actually enhances the success rate of future r and d products as well as ingredients because if you know how certain ingredients are used in the market and how, um, um, how successful new ingredient launches are and how [00:06:00] successful the pro uh beauty product launches are, you, if you can link everything together, the, uh, the r and d will be more successful.
Deanna: Can you say more about that? Can you gimme like a for instance.
Karen: Absolutely. Um, one of the metrics we track is, uh, when a product, when a new product is launched, um, we track the success of the product use, uh, using multiple proxies. So, um, we, we aggregate the consumer reviews and all these social signals across different, uh, retailers and platforms. So. It is not just tracking new product launches.
We also track how successful these products are, and we decipher what ingredients and ingredients blends are used in these, uh, products. And we link it back to the suppliers. And we, uh, we also display the, uh, contrast manufacturers and brands in the middle. So if you are able to link the entire value chain, then suddenly you can see the linking linkages between your ingredient.
How is that impacting the success of [00:07:00] the product? And, um, lots of new products are launched in the market, but many of them fail. But if you are able to find the patterns in why, what is making, what successful in what way, and then that is a very huge, uh, is a huge value add to the industry. So.
Deanna: Absolutely. And it makes me think about the challenge that so many, uh, especially new ingredient makers have. Um. Almost a desire to put their trade name on the, you know, the consumer package in hopes that the consumer will start to, um, you know, request a very specific version of an ingredient that's available.
And even without sort of revealing the trade name or, or using that through to the end consumer, your technology is, is in a way serving that purpose. It's creating a, a, a direct through line from consumer. Uh, benefits, uh, reviews, appreciation, if you will, to a particular ingredient maker, isn't it?
Karen: Absolutely.
Deanna: Yeah, so interesting.
So, um, I'm curious [00:08:00] now, um, maybe to think about how accessible technology like yours, is it, I'm imagining it's, you know, not, not perfectly affordable for every brand in the market, but who, who can work with this kind of technology? Is it only major multinationals at this point, or is it, is it a bit more accessible?
Karen: We had a, um, a customer conversation yesterday. Um, they are from a, um, uh, UK based manufacturer of. Chemicals. They said they, they actually stopped me in the middle of the demo and say. Uh, I have seen many platforms. So for platforms that have very accurate information like yours and super granular data, usually it's expand. Very expensive,
actually. Um, I told him, uh, the price, it is very accessible and it is, uh, because, uh, now the world is also changing. So, uh, we are AI first company, so we only have a handful of employees here.
Compare with the, uh, the last [00:09:00] generation industry intelligence companies where you need a lot of developers, you need a lot of, um, um, industry professionals labeling each
bit of the data.
So, okay, someone needs to label, okay, this, uh, chemical is a effect. That chemical is an active. So it is a very, uh, different world now. So we are able to ex completely streamline our operation, um, and we bring a lot of value back to our customers.
So we are completely democratizing industry intelligence for this industry.
Deanna: Mm-hmm. Mm-hmm. No, this is, this is what people want from technology. It's it, um, yeah. It's nice to hear we're moving in that direction. Um. I'm curious to know too, when the sort of, um, data-driven transparency at, at the level that you're describing, right? When does that start to threaten intellectual property or IP for suppliers?
Are there limits in place? You know, do there need to be safeguards? Tell, tell me about that aspect of this.
Karen: Ah, [00:10:00] actually, um, potentially in the future it'll be. Um, because of AI and because of the superpower of ai, uh, in terms of, um, digesting and reasoning with data sets in the future, um, around ip. So for example, if it'll be possible, for example, uh, on our product pipeline in the future, um, we, we are looking into patent intelligence actually. So in the future, for example. Um, it'll be very visible when a major ingredient supplier, some of the patents are expiring,
and then would, um, uh, encourage, uh, other competitors or even their, uh, their own customers to. To use similar technologies that used to be proprietary to them. So I think it will at the same time democratize the, uh, the information and data, uh, space, but at the same [00:11:00] time, it'll become more visible.
Who is filing what patents for, for example. Um. Say our RNG Innovation Center, what recent patents have they filed? And then, uh, then it becomes very visible to others who are tracking these aspects of things. So, um, I think in, in that direction, there will be a lot of impact.
Deanna: Yeah. Yeah. No, it is so interesting to think about and certainly that's, you know, as you've said or suggested, perhaps all of this information is information that. Brands and suppliers have been looking for in the past. Right. And, and the same is true with intellectual property and, and patent data. Um, uh, we certainly see folks, um, you know, borrowing technologies or, or you know, um, trying to create their own version of patented technologies even without a platform like yours.
It's interesting to think about how things will evolve there.
Karen: I
think it'll accelerate it a lot. Uh, I think you can already do [00:12:00] a lot of things today if you have a large team of people doing a lot of manual work and trying to maybe infringe other people's patents or just take other people's expired patents and experiment with those. But in the future with ai, uh, it'll be much, much, much faster and more competitive.
Deanna: Mm-hmm. Mm-hmm. I, I think I've asked this of, of other folks in the, in the AI space, but I, I do wonder if this opens up an opportunity or maybe encourages, um, more creativity and, and. Innovation from brands and suppliers in, in an effort to really differentiate themselves in the market. Do you see that or do you see it's, you know, more likely, you know, we talked earlier about if a particular ingredient is really delivering benefits, is everyone going to want to use the same ingredient, or is there more space here for, um, for variety?
I wonder.
Karen: Um, this is, uh, an amazing, uh, point because actually this is the point. I'm not sure whether it is all positive. So, um, for example, we were observing, um, the ingredient [00:13:00] trends in the market six months ago. A high-end beauty brand launched a really, really expensive serum with a unique root extract in
it. Within weeks, um, our system already tracks lots of copycats, uh, or like similar products in the market,
um, with a lower concentration of the root extract and packaged in different ways. So, um, I, I think the market is. Moving very, very fast. It is even faster than before. So in the old days, it takes, uh, two to three years to develop a new product. And in recent times it takes several months, uh, with the testing, with the validation, re, re and, and nowadays I think you can create a new product, um, that are similar to an existing product, um, in weeks.
Um. I'm not always, I, I'm not sure this is always positive because, um, it takes a lot of r and d and, and effort
to create something innovative and successful.
And, um, I think this is, [00:14:00] um, something the brands and the contract manufacturers, ingredient manufacturers have to have to have to work with.
Deanna: No, and, and to be fair, there have always been, you know, again, those activities going on as well in our industry, it's, it's not the technology, it's the, it's the people who choose to, to use it in, in various ways. But, um, let's think about some of the, the positive advantages here as well. I would like to think that an AI platform like this can actually help build trust, uh, throughout the industry value chain.
I'm wondering if you're hearing about any interesting changes in that way. Do you, do you see companies finding new ways to collaborate or, or, or such because of this kind of technology? I.
Karen: Absolutely because, um, um, the industry is becoming more and more transparent and, um, um. If, if, um, a company, um, creates a very successful product, no matter it is a beauty product, or if they created a fast growing brand, so suddenly because we track the geographical reach [00:15:00] of brands and products, suddenly this is a upshot and suddenly they're gaining lots of shares in different platforms. Um, and then, um, um, and also for, uh, ingredient suppliers and manufacturers, they come up. They, they, they. Spends a lot of, um, uh, effort in creating a very effective new ingredient, and then suddenly it takes off. So, um, by making the industry transparent, these success stories, uh, would attract a lot of, um, uh, partners and, um, new customers. So it actually encourages, um, um, uh, new collaboration. Between teams, um, and then because people want to, to, to, to do new things that are more, um, better and better compared to the old things. So, um, I, I, I'm seeing, um, a lot of positive things out coming out of these, uh, um, making success stories now.
Deanna: Yeah. Yeah. Oh no, that's interesting to think about. Partnership opportunities coming. I mean, it makes good sense, but um, yeah, that's very cool. I [00:16:00] do want to, um, I'm starting to ask some of our guests a little bit about the behind the scenes at their companies, and I definitely wanna hear a little bit about, um, the tech you're using in-house.
Um, but before I do that, I just wanna give you the chance. Is there anything else you think that ingredient makers or contract manufacturers or brands should know about this kind of AI platform? Is there something I didn't ask you?
Karen: Maybe our upcoming roadmap, but that is a separate story.
Deanna: Yeah, if, if there's something in the future that, that we might be watching for, that would be helpful.
Karen: Hmm. For example, we are launching a, uh, uh, private dashboard feature, which is a completely, um, confidential and private feature for ingredient manufacturers to upload the. Ingredients portfolio so that they can have a bird eye view of the entire market demand and, uh, dynamics of their p uh, of the ingredients.
So who is by what, where, which area is growing, uh, which new applications are using and old ingredients. And, um, and then we are, um, [00:17:00] creating a lot of new things as such as, uh, patent intelligence, um, later this year. And, uh, new features such as, um, um, allowing people to. Um, search for certain aspects of a beauty product in consumer reviews.
So if you want to analyze, um, the difference between, uh, any, any customer feedbacks on synthetic fragrance versus, uh, natural fragrance such as, uh, seed oil or different, uh, botanic oils, then you can really easily mine the data using ai,
um, on our platform.
Deanna: No, that's helpful. Thank you for that. Um, you know, so as I suggested, you know, um, not every company in the cosmetics and personal care industry is necessarily an early adopter when it comes to, uh, new digital technologies. Um, but certainly a company like yours is very future forward. And I do wanna help think about your approach to technologies.
I am still learning about ai, of course. But it seems to me that you're working very heavily with what we might think of as predictive [00:18:00] ai, the sort of, uh, data-based forecast technology. But I'm wondering about some of these other types of artificial intelligence, like cognitive ai, that, as I understand it, stimulates human thought, intuitive ai that can kind of infer meaning, um, or even other kinds of ai.
Can you, can you tell me the, what's going on in terms of AI at gravel ai?
Karen: Um, okay. So actually at the company level, we. A lot of our code is written by ai, so 70% of our company code is written by ai. So it is, it is, this is really significant. This is one of the major reasons why we can democratize, uh, industry intelligence at a lowish cost. Um, from data collection to building the data pipeline to a data ingestion and a synthesis of the data, and then creating the, um, the interactive experiences that our customers see. 70% or more of the code is written by AI right
Deanna: Yeah. And if, [00:19:00] if something is like written or created, it's, that's generative ai, that's a
Karen: Generated by ai. Yes.
And at the same time, on the commercial side and on the operations side, um, we use AI agents internally. So a lot of the key tasks, uh, from account management to, uh, like, uh, some of the customer success and also, uh, when people book demo and how I. Uh, sometimes our customers wonder how I can respond so quickly. Uh, within, uh, two seconds. Uh, we, I already say, okay, I look forward to meeting you. And there are multiple interactions that I have, uh, we, we have automated at a company level. I.
Deanna: I'm a little jealous. My email could use some help like that. Um, yeah, I, you know, and I ask these questions to help companies really start to think about the practical applications and, and why, um, you know, they might, you know, also bring in technologies like this and, and how it can maybe free up, um, time for their team in other ways.
Are, are there any other notes you would have about, you know, companies considering [00:20:00] automation technologies or AI technologies like this?
Karen: Um. I have a, um, a thought in order to automate, I think, um, we need to identify certain key workflows that can be standardized. So once we have standardized those workflows, um, there are many different tools. So you can combine those tools to automate each step of the workflows. Uh, I I, but the standardization in the beginning is very important.
Deanna: Mm-hmm. No, that makes, that makes good sense. This is all very helpful to know. I do wanna mention for our listeners, um, if they're especially curious about AI in beauty, you can go ahead and scroll back through some previous episodes, uh, from the Cosmo factory. I have interviews there with, uh, say, avision, a company that uses AI for quality control in manufacturing.
Um, I spoke with Cambium where they're using AI for ingredient development. Potion ai, um, is a platform that, uh, helps with product formulation support. Perfect Corp. I spoke with as well. [00:21:00] They bring AI tech for consumer engagement. Of course, all of these conversations are available in the Cosmo factory archive.
And now, uh, gravel AI is on that list as well. Karen, you have so much enthusiasm about what new digital tech can do for our industry. I, I thank you so much for being a guest on the Cosmo Factory Podcast.
Karen: Thank you so much.