NotCo: AI technology enhances fragrance formulation “in seconds”
04 Dec 2024 | NotCo
NotCo is branching out of the food industry and into fragrance formulation. NotCo CTO Karim Pichara speaks to Personal Care Insights about its Generative AI Fragrance Formulator, made in collaboration with Cramer, a Latin American flavor and fragrance company. Using NotCo’s AI technology, the formulator can craft bespoke fragrances in seconds.
This is Sabina Waldeck for Personal Care Insights.
Joining me today is Naco's CTO and co-founder, Karim Picarra.
Welcome, Karim.
Thank you, Sabine.
Thanks for having me here.
All right, so let's just jump right into it.
Could you describe in brief how the not cogenerative AI fragrance formulator works?
OK, great.
No, you know, we, in our early days at Norco, we realized that there was a huge opportunity whenever we needed to hit this phase of product creation and where we needed the, the fragrance or, or the flavor for the product was something totally.
Unknown for me in particular because before I didn't work in the food space ever like I I'm coming from the AI world but this was the first time I was getting into the the food tech space, right?
And, and I say wow this is crazy.
We had to wait to wait like.
5 to 6 months in order to get to the right flavor or fragrance on every single product and I we understood that there was there was a huge opportunity why?
Because there wasn't any technology involved for the creation or formulation and was only relying on the expertise of these flavor or fragrance chemists that they have 20 years of experience at our various cars and.
Very, very busy, right, in general.
And at the same time, you know, that they would create a formulation, send, send to you the bottle, then you try to say, you know, I don't like it so much, you send it back.
He has to come, create a new formulation, compound it again.
It was crazy.
So we, finally, we could, you know, partner up with a, with a Fragrance's house in where we could access their, their data.
So we could build an AI that, that was trained with all this data to understand the connection between the molecular structure of, of all the chemistry behind the fragrances.
And you know, the sensorial description in words of those fragrances.
So with that model, we, when we try, after we training and we tried, we realized that we could formulate in seconds something that we, it could take like a few days for a person and at the same time, you know, it allowed you to iterate super quickly, like basically, you know, you try a formulation, you compound it, maybe you like out of these 10 formulas, you like these 3, you select, you ask for another 7, you know.
It's so, it is streamlining the entire process of fragrance creation like exponentially.
So that's the way, that's the way it works.
Right.
And so, how can it help make, so you're explaining, yeah, how it can make fragrances quicker, when it comes to sort of like end products, how does it make the whole sort of system quicker?
That's a good point.
Yeah, there's not, it's not just about speed.
There's another very important factor which is innovation.
And novelty, because in general when you have, you know, people working and even if, even worse, if they have 20 years of experience, they are totally biased towards the formulations they have in their head that worked before.
So let's say like if they, they are creating a smell of or, or a fragrance of, of a certain fruit or like a strawberry, maybe they created 150 strawberry fragrances before.
So, of course, they're gonna go back and whenever we need to create this, if they need to create this fragrance again, they will definitely be 100% biased for, of their prior formulations.
They will never try any, any necessary new combination of, of ingredients, right?
So this technology is letting people.
To balance between the previous knowledge and at the same time allow for creativity by creating new combinations that maybe will bring new notes, right?
And the good part of it is like it is all in, in the, in your hands because when you type or you prompt.
The sensations you're gonna get from, from this fragrance, the model understands your perception, understands what you want, but at the same time understands that there is a baseline that needs to accomplish in order to have the, the, that sensorial profile.
So, the AI is able to balance both and bring you the previous knowledge to gain speed.
And at the same time bring you the possibility to create new things.
So that's the, the, the main difference.
It's, it's not just about speed, but to your, to your point about exactly how faster it is, yeah, you, you can tell that at the end.
If you, you might take like 3 or 4 weeks in designing a fragrance.
Assuming you don't have any, any, delays on sending fragrances back and forth like in a perfect and perfect scenario, then with, with this technology, you can take 2 days, right?
Right, and so because it is like becoming more streamlined, will it take away any sort of jobs?
Cause that's a big concern with AI and it like making things easier, is that it potentially could take jobs away from people.
So in this situation, will that happen?
Great question.
Yeah, I don't see technology taking away jobs.
I just see AI technologies augmenting people's job.
So to me it's like, of course, companies are not going to fire their flavor experts or fragrance experts.
They are super scars.
They need them and, and still this technology.
It is relying on, on people's the, the, the control.
The way you design a fragrance, it depends on the person that is typing, you know, iterating, sending this feedback to the technology.
The beauty of this is that, if you have two people working in parallel, designing the same fragrance with the same technology, they will end up in different results, right?
So that proves that the human input is super important.
Is it, and the AI is, is here just to augment people's work.
So I see this of instead of if you had one person being able to work in one project at a time with this technology, maybe they can work in 10 projects.
So I don't see people, losing their jobs.
I see people being way more productive and taking way more challenges than before thanks to AI.
Right, OK, that's really interesting.
And so how would you say that AI in general is influencing fragrances and the fragrance world and fragrance formulation, and how do you expect it to change and grow in the future?
The influence, first of all, I, I believe that now we could create novel fragrances that that could be connected with, with emotions, with memories, and.
With way more abstracted abstract concepts because as far as we can express those concepts in words then the AI can help in bringing the fragrance that resembles sometimes like somehow that experience so I believe that we're gonna experience novel fragrances that's one thing.
Also, I believe that with time we're gonna be able to have more personalized fragrances as in where maybe we would be able to have com to companies will be able to compound formulations and.
More tailor-made for people.
That's another thing I see happening in the future with all the digitalization.
And you know, in general, acceleration brings new, new discoveries like it might be tangential to, to fragrances.
If you streamline the process and you allow these big companies to focus in other thing, in other aspects, they, they would be able to.
Do refocus their research, maybe, do way more with less, bring novel solutions in every aspects connected to fragrances as , and, and I also feel that this will impact the, the food space, not just personal care, home care, but at the same time, you know, the food space with novel flavors because they are super related.
Right, and for people who might not be as kind of aware on how AI works and things like that, could you give kind of an example of this sort of future looking perspective on something that can come from a memory or something personalized?
Could you give kind of like an anecdote of what it could look like?
Yeah, of course you know that AI today is, is known that AI understands very language, right?
Like from all the texts that we've been sharing.
From all the texts that we've been sharing on, on internet, AI is really, really good at understanding English or any other language.
So there is data that is expressing the words that describe.
Sense fragrances, experiences, and those words are, part of a, a more specific jargon, right?
So the, the, the words used to describe sense are more technical words and of course they are not freely available up there online in the same amounts that normal language, right?
So when you are able to gather a data sets, the data sets that bring the connection between those new words and the molecular structure of the formulations, then you extend the knowledge of the AI by training it over this new information.
So when the AI gets exposed to this new data that is telling you, you know, these words describe this formula and create this sensorial profile.
Then the AI start making all those connections and understanding what is how, how to map between these formulas to this.
Sensorial profile.
It needs to understand che chemistry and words, and then putting all that together, so with the right data, you get that effect at the end.
OK, yeah, no, that's super interesting.
I think that was a really good explanation.
And do you have anything else to add that you feel like we haven't covered today?
No, I think that, we are super happy to, to tap into this, space like, as a food tech company, we were not expected, expecting to, to get into the france or personal care, but it's, is, this is a great example of what I was mentioning before about sometimes when you develop great technology and you start streamlining other some processes, then you realize that your technology or your discoveries might apply to different worlds and then.
Everything start, you know, getting connected and you can create.
A better and stronger impact out there in different industries.
So, so it's super exciting and, and yeah, just waiting to have for us to have maybe the, the new upcoming fragrances available for, for everyone, yeah.
Awesome.
Yeah, thank you so much for speaking with me and Karin, thank you for joining Personal Care Insights.
Thank you, thanks.
