The hunt for perfume perfection could be over thanks to this Illawarra mathematician

Imagine you walk into a store, answer a few questions about your mood and preferences and before your very eyes, as if by magic, a perfume is mixed with precision for you alone – and it’s perfection.

This is the future one Illawarra-born and bred woman is delivering – though the “magic” she employs is artificial intelligence (AI).

Katie Buchhorn has launched Signature Scent AI, an AI-driven service matching customers to their ideal perfume.

“There’s a relationship between your other sensory preferences – your taste, vision, what environments you like to spend time in – and what you like to smell. It’s a really complex relationship, but the key to understanding this is about asking the right kind of questions,” she says.

“If we can capture the right kind of input data and put it into models, we could be the first to gain a real understanding of it.”

It’s one of those ideas that doesn’t seem possible – until it is. But an accurate predictive model requires enormous amounts of data and an effective collection method.

Between the rise of “big data” and Katie’s PhD in Bayesian statistics and machine learning, however, new possibilities are here.

Though first established in the late 18th century, Bayesian statistics has found new life in the age of information, says Katie.

“There has long been an academic divide between ‘frequentism’ – the statistical methods commonly taught in school – and Bayesian statistics, a philosophy that helps us understand the inherent randomness of our world,” Katie says.

“Now we’ve had a big boom of data, and we need to be able to process and understand this data like we’ve never needed before. We’re learning more about our world through Bayesian statistics, which takes into consideration uncertainty in a principled and thorough way. It’s a fundamental paradigm shift in the consideration of variables.”

The seed for Signature Scent AI was planted some years ago, though initially it was more about addressing a gap in the scent market for natural products.

“Our skin is our largest organ and perfume is something we carry with us all day, yet there’s very little transparency around the formulas,” Katie says.

“For instance, perfumers can use natural rose or synthesise the smell with a cocktail of harmful chemicals to mimic it, and there are no strict legislative requirements to disclose that.

“Selection is another challenge. I have tested scents on hundreds of people, asking them to try eight perfumes, and have found by the fifth or sixth they start to struggle with sensory overload.

“I thought, ‘What if we could create a solution using technology and intelligence far beyond anything we’ve ever known, but is now within our reach?'”

Using principles from Bayesian statistics to formulate an algorithm that can create a match to something as deeply personal and, at times, ephemeral as a person’s preferred scent is an iterative process.

While Katie’s questionnaire and model will be refined over time, she’s ready to go to market with the latest.

Customers answer a brief questionnaire designed to uncover their unique sensory preferences in under 90 seconds, then rate a sample pack of five perfumes. Using cutting-edge AI, that data yields the signature scent that will be delivered to their door.

“We’re entering a new age of product personalisation, and that’s what I get excited about,” Katie says. “I think the big wave of AI-permeation is fast approaching, and I don’t just want to watch it — I want to ride it.”

You can find out your Signature Scent now.