It is 4AM on Friday in Hong Kong, and as I am finally preparing to go to bed, I scan the Haut.AI QR code. The Skin SaaS system instantly pops up. I position my face within the oval space on the Google Pixel, but just before I snap a picture, the system warns me that the lighting conditions are far from perfect. I head to the bathroom in search of a better light source. When the system scans my face, I learn that my perceived age is 49. That stings a little because, chronologically, I am 43, and I have dedicated my life to advancing longevity science. But there is a good reason for this poor result: my schedule over the past two weeks has been insane.I’ve averaged just 3.5–4 hours of sleep each night and certainly did not get enough exercise. In other words, I look like a zombie.
The system analyzes my skin further using the latest and greatest in next-generation artificial intelligence technology to provide me with other valuable insights about my skin. After spending a month in the Middle East, it comes as no surprise that my skin is rather dry and needs a good moisturizer. I follow the app’s recommendations, set my alarm, and finally go to sleep. I wake up at 10 AM on Saturday feeling human again. Those five hours have been very invigorating, so I follow the procedure once more. Now, my perceived age is 43 years old, but my translucency index, that elusive feature that makes our skin look young and healthy, is poor. I grab my favorite Nivea Men cream, apply it generously to my face and body, and set out for the gym.
Unlike my dear friends who see me at conferences, the Haut.AI Skin SaaS system does not lie. It was trained on millions of carefully annotated images of skin, collected from both high-resolution equipment and selfies from all over the world. In fact, this advanced platform is used by most of the top skincare brands that perform serious R&D on their products. I’ve had access to it since I joined Haut.AI as an advisor. Right now, this innovative system is available only to potential customers and R&D partners; it’s not in the hands of consumers yet.
On my return from the gym, I wait for the clock to strike 9 AM in Tallinn, Estonia, then call Anastasia Georgievskaya, the founder and CEO of Haut.AI.
“Guten Morgen, meine liebe Geschaftsführerin”, I greet her in German. “Kuidas teil läheb, professor?”, she replies jokingly as it turned out, in Estonian, then we switch to English. I report on my disappointing results, and we have a lengthy discussion about the algorithms, as Haut.AI is currently piloting several new generative approaches. The latest is a highly ambitious digital skin twin simulation that allows the company to model entire human populations in an unlimited diversity of scenarios. That’s the images of billions of people, almost one hundred variable factors, and hundreds of thousands of skin products input into a computer simulation to achieve outstanding accuracy and results—amazing!
Anastasia informs me that she has just published her first research paper as a single author, meaning that she has done all of the experiments by herself, which is very rare in multidisciplinary sciences these days. We talk about Haut.AI’s recent mega-collaboration with Ulta Beauty, the largest beauty retailer in the US, and the many scientific challenges they’ve just solved. It’s encouraging that some beauty retailers are truly heavily engaged in R&D and are not just trying to push their products to customers. We also discuss the research papers that her team is working on and the possible applications of her new algorithms, specifically the potential for new target discovery and linking biological omics with imaging data. Our conversation moves to the robotics research lab I am building at Insilco, and she offers some advice on how to connect imaging, epigenetic, and expression data using one of the approaches they are trying at Haut.AI.
Finally, Anastasia gives me an update on the company’s operations, and I am pleasantly surprised to hear that they are profitable, have several million in revenue, and the new SaaS platform is being rapidly adopted by the many small and medium online retailers that want to use innovative AI-guided skincare to differentiate. She wonders if they should raise more money from investors to introduce new products or if the capital markets are so bad right now that it is better to rely on their own internally generated funds and focus on the science. Her hobby is managerial accounting, something that I try to stay away from, and she throws a ton of figures at me, explaining that they can grow an additional 50–70% without external capital, even if the market is bad.“Consumers are becoming much more conscious about what they buy, and they want to pay for results, not for expensive brands and marketing. That is where we do not have much competition—we lead by doing good science and providing consistent and robust solutions. And validating this tech is not easy, so I’m not afraid of someone out-competing us. On the contrary, we often help our competitors to ensure that the industry does not get a bad reputation,” she tells me.
I’m quietly impressed. “I have known this young lady since she was a twenty-three-year-old master’s student,” I think to myself.
You see, I first met Anastasia in 2015 when she briefly joined Insilico as my assistant. The first interview did not go well, and I hesitated. But she was introduced by another young scientist, who had provided phenomenal references. After just half a year on the job, it became clear that she was much more advanced and entrepreneurial than any of us had expected. She wanted to dive into deep learning and found herself a wonderful coach in the brilliant engineer Konstantin Kiselev. At that time, Konstantin worked as a consultant for NVIDIA, evangelizing and teaching CUDA GPU technology to the many emerging AI and gaming companies of the time. Together, they decided to start developing all kinds of apps to explore human skincare and beauty. They started with RYNKL, a now-defunct app that could annotate facial wrinkles and explain where it may be necessary to add some fillers. They quickly realized that age is the main determinant of wrinkles and also of beauty. That led them to the renowned Beauty.AI competition, where machine learning scientists competed for prizes using AI to evaluate human beauty.
Having gained substantial experience in the application of AI in skin care, they moved to Estonia in 2018 and formed Haut.AI, working as a contract research organization for some of the world’s largest skincare brands. And some of these brands have amazing R&D. For example, one of the reasons I now use Nivea cream is because I saw their R&D center in Hamburg—the research campus is the size of a modest university! Unlike other skincare conglomerates, like L’Oreal, Beiersdorf focuses on two brands: Nivea and Eucerin. And Nivea is hugely popular, which means they cannot screw up. A small error somewhere could lead to dramatic consequences. That is why every product that Nivea makes is carefully studied and analyzed by teams of high-caliber German scientists. When deep learning technologies started to outperform humans in image recognition, they decided to apply these technologies to the massive research data sets they had collected over many decades. And at that time, Haut.AI was already the best technology around.
I can only imagine how Anastasia, just twenty-six years old, felt presenting her work to this group of ultra-elite German skin scientists, but they quickly established an R&D partnership and Haut.AI started developing AI technologies together with them, scouting the entire world for talent and innovation in skincare AI.
Other brands also quickly realized the value of Haut.AI. They could send a bunch of anonymized skin images to the cloud and get back high-quality annotations with hundreds of predicted features or massive amounts of synthetic, AI-generated “deep fake” skin data.
Over the course of two years, Haut.AI managed to create a database of pretty much every skincare ingredient on the planet, as well as almost every cosmetic product. Working very closely with brands in India, Asia, and Africa, they ensured that the platform’s algorithms could work effectively with any skin tone, skin color, weather conditions and health status, and offer solutions at any income level. Haut.AI turned into a one-stop shop for product recommendations, pairing any skin type with the product most likely to result in quantifiable improvement.
Today, Anastasia is twenty-nine years old. She is running a global skincare AI powerhouse with over 30 employees in Estonia, other EU countries, and the US. Her goal is to make high-quality skin care accessible, scientifically-backed and focused on prevention.
Before the test of Haut.AI’s new SaaS demo platform, I took the liberty of asking Anastasia a few questions:
Alex: Few people know Haut.AI as a longevity company, but for the past five years, every year, we meet at the ARDD conferences, first in Basel and now in Copenhagen. Can you please describe your interest in longevity and how Haut.AI fits into the ecosystem?
Anastasia: I should start by saying what probably every research paper about skin opens with: the skin is a very complex system. It’s the barrier that separates us from the entire world, but it is also perceived as a reflection of our health and wellness. The fact that skin condition is affected by both extrinsic and intrinsic factors, as well as natural aging processes, makes the skin a perfect aging model.
If you think of skin from a practical aging research perspective, it is relatively easy to sample. Not many people will agree to a biopsy, but the range of tools available to study the skin is just enormous. You can do omics, genetics tests, swabs, tape strippings, biophysical measurements, and more. You will most likely only need to do it once, or from time to time perhaps. But, what about selfies? If a person can post a selfie on their social media at least once a week, then they can take a photo of sufficient quality to study their skin. Just think about it: you can generate multiple data points about your health and aging pace with something as simple and ubiquitous as a selfie. That’s incredible.
Alex: Haut.AI is now a global leader in AI for skin care, and all of the top brands doing their own R&D are working with you. What are the key factors for this success, and how did you manage to achieve this out of Estonia?
Anastasia: First, one thing that makes Estonian startups so good at the export of IT services is that they never assume they can just stay in Estonia. Unlike the US, China or UK markets, Estonia is pretty small. This is why founders in Estonia think big and act bold from the very beginning.
Second, I credit our success to the high standards we have always celebrated in our company. We are always trying to use novel techniques and new frameworks, and we familiarize ourselves with the latest research papers so we know every tiny detail and the limitations of every algorithm. It also helps that our founders and first employees had scientific backgrounds. Science teaches people to be curious and to always ask “why?” We ask “why?” and “what if?” all the time.
Lastly, HautAI is an IT company for skincare and longevity. Many beauty companies are trying to be IT companies. But can we say a marketplace with an AI component, like a chatbot or a mobile app, makes it an IT company? I’d say it does not. At Haut.AI, our development is executed in two-week sprints. We update our software twice a month. This year, we have already had 27 releases and delivered hundreds of new features.
Having great flexibility and agility in our product development has helped us to react very quickly to fast-changing consumer trends. This gives brands working with us a competitive advantage, allowing them to keep up with the pace of skincare digitalization and even predict consumer needs so they can be proactive.
Alex: The US is a major market for beauty products, and you are already working with Ulta Beauty. Why not move the company to the US entirely?
Anastasia: We started in Estonia as it is just an amazing country for innovation. Everything is digital, the startup community is incredible, and the weather is rainy, so you can focus fully on work.
Jokes aside, our journey started in Estonia, but as we work globally (in 32 countries) and the US is such a huge market, at some point, we will establish a US branch.
Alex: What can you tell me about your business growth plan and fundraising? My last article on Marble generated massive interest in the VC community, and many investors reached out to the wonderful Dr. Milanova. What would you do if you started getting investor inquiries?
Anastasia: That interview was great, by the way! When it comes to Haut.AI, we have been self-sustainable and revenue generating for some time. We have built a product, we serve multiple clients, and as a result, we get great interest and demand for our solution. It is time for us to grow aggressively and explore new business models and R&D ideas. We want to expand in edge calculations, work deeper on the research of multi-dimensional skin data, and, most importantly, continue delivering our product as a highly scalable and robust solution. That being said, I think we will start fundraising very soon. We are not afraid of a turbulent market, as our solution is not a “vitamin”—it is an antibiotic, which means it will always be useful. We have our first research finalized, which proves that we can track individual treatments and the effects of skincare products, as well as answer the eternal question: “what I should apply on my face today to look better?!”
Alex: I have started using your experimental Haut.AI Skin SaaS and it told me that I look older than I am. And, also, my skin is pretty crappy. Can you recommend a few simple ways for me to improve my skin? But you can not recommend sleep, diet, meditation, or avoiding UV—only skin products and procedures.
Sure, I have some tips for you. Let me put on my “influencer” hat. You should start by using proper skin cleansing every morning after you wake up and before bed. I personally use non-foaming products, as I believe surfactants (foaming agents) remove too much skin sebum, which leads to a lipid coat imbalance and a decrease in the skin’s protective function. But you can opt for any other type of cleansing product. Just clean your skin every single day, twice a day, for at least 45 days. Skin renewal cycle typically takes between 21 and 60 days, and because you don’t know what stage of the cycle you’re at when you start cleaning your skin, you need to do it for at least 45 days. Generally, try everything for at least 45 consecutive days if you want to see results.
You scored not so good on pores metrics, but that’s ok. You need to exfoliate the rough skin on the outer layer. You would benefit from home acid peelings, or maybe do high-concentration retinol once in a while with a professional.
To improve your skin’s elasticity and sagging, you can try massage. The benefits here come from muscle relaxation. To improve skin matrix structure you can try professional treatments, like Intense Pulsed Light.
You know, I’m asked this question a lot. Maybe I should write a book about it!
Alex: What does this mean for the everyday person or consumer, and what future are you hoping to create with Haut.AI?
Let’s think about the typical skincare experience for an everyday consumer. Let’s say their name is Lucy. Lucy wants to shop for skincare online, or she walks into Ulta or Sephora store. The number and variety of products are enormous and overwhelming. How can she choose suitable products that will improve her concerns? What if she also wants to work on her skin resilience for better skin health long term? She asks these questions every time she walks into a store because our skin health and look are important for our self-perception and how others see us. Does it make sense?
Alex: Yes. Please, go on…
Anastasia: Lucy takes her phone and uses Haut.AI app that explains to her what her skin needs are today and what products will enable her to build skin resilience. Lucy could end up with a set of products worth 150$, but half of them are wrong for her. Another scenario is that with the same budget, Lucy would use AI to pick the products that her skin needs and will benefit the most and do it fast and frustration-free. Lucy got suitable products, and she will see the positive change. She understands her skin and knows how she can manage it. When she comes back to the store that offered her a pleasant shopping experience with AI, the system will refine her recommendations to focus on the next skin goal. This is Lucy’s story.
Let’s also talk about the fact that the skincare field is sadly not sustainable. 70% of plastic packaging goes to landfill, and there is a lot of plastic in beauty packaging. At least 10% of beauty products are discarded because of overproduction and imperfect logistics. How do we shift from where we are now towards sustainability? One of the answers is personalisation. We need to start retargeting existing products while critically evaluating what consumers really need. You cannot only rely on consumers’ responses in quizzes or surveys. Obviously, consumers are biased and can be both too critical or unaware of their skin conditions. The beauty industry should rely on data generated by technology like Haut.AI to explore different markets and understand real consumer demands for products. Personalisation is the key trend and a new beauty paradigm that benefits all stakeholders.
Alex: Can you tell me about your latest achievements? For example, this month?
October was quite eventful. First, we were named Skincare Aesthetics Disruptor by the latest Longevity Technology report.
We announced our partnership with Breezometer, an environmental intelligence company, and we work on augmenting skincare recommendation engines with environmental data. By the way, Breezometer was recently acquired by Google, and we are very happy for them.
I published my first single-author research paper in the Journal of Plastic and Reconstructive Surgery. In this work, I discuss how incorporating data from different population groups, and visual biomarkers can make skin health evaluations more accurate.
Haut.AI Skin SaaS platform released a new product, Skin App Constructor, which allows companies of any size set up their own branded skin scan application in less than 3 hours. When our customers see Skin App Constructor now, they say that is magic.
After the interview, we perform the test of Anastasia’s face using one of the Haut.AI demo tools for clients demonstrating its Software as a Service toolkit that the customers can integrate into their research and commercial systems.