Predicting the Next Viral Food Trends
Co-founder and CEO of Ai Palette, Somsubhra GanChoudhuri, shares how artificial intelligence and machine learning are used to help companies identify and predict upcoming food trends.
Peanut butter beer – this may or may not sound appetising to you. However, artificial intelligence (AI) technology has shown that this might just be the town’s next “in” flavour.
Ai Palette, a local foodtech startup, developed a predictive analytics platform that can identify and analyse food trends in a particular market to determine what makes a certain ingredient or flavour popular and even predict how long that growth might last.
The platform can also predict what types of food or ingredients are up and coming, which can help companies create products to meet the latest needs of consumers.
“For instance, consumers recently started looking at how to boost their gut health. Companies using Ai Palette will then be able to see and understand this trend in real-time, and its predictive analysis can help them make product decisions with confidence,” Ai Palette’s co-founder and CEO, Somsubhra GanChoudhuri shared.
Prior to starting Ai Palette in 2018, GanChoudhuri had worked in the food industry for over seven years, where he saw how new food products are traditionally created.
“If we were to create a new coffee flavour for the Indonesian market, we would have to fly down to Jakarta, visit the cafes and supermarkets, taste the coffee served there or see what kind of products are available, write down our notes, before coming to a decision or recommendation.”
Co-founders of Ai Palette - Himanshu Upreti (L) and Somsubhra GanChoudhuri.
These market visits took up a lot of time and costs, and often included the biases of people conducting the visits. “About 85 per cent to 90 per cent of new innovations fail in the industry, which is a very high rate of failure,” shared GanChoudhuri.
While mulling over ways to improve such innovative approaches, GanChoudhuri noticed there is already a lot of data available publicly online.
Identifying New Food Trends
Together with his co-founder Himanshu Upreti, who has a background in big data analytics and advanced machine learning, they worked on using AI to draw meaningful correlations between these large amounts of data points.
The platform currently collects data from multiple sources such as social media, restaurant menus, online recipes and retail products available on search engines. Natural Language Processing (NLP) algorithms are then used to decipher the data before using AI and machine learning to identify trends and predict a food product that may go viral.
So how does Ai Palette identify a trend that has not appeared yet?
GanChoudhuri explained that their system uses the named-entity recognition in AI to identify new words or terms in the data pool. The algorithm then tracks to ensure that there is a significant volume of such data points to classify it as a trend. For example, “dalgona coffee” became a viral hit on social media last year, when most people would not have heard of the word “dalgona” before.
“There is a threshold for every market that the data has to cross before we can say with confidence that this is a new trend. This includes the number of data points around the particular trend, how it has evolved in the past, and how it is manifesting itself in menus and other products today.”
Supports 15 Languages
Ai Palette’s ability to support multiple languages is also one of its most significant differentiating factors from other trend prediction AI platforms. It currently supports 15 languages, including Asian languages like Mandarin, Japanese, Vietnamese, and European languages like Spanish, French and German, with plans to add even more.
GanChoudhuri said that as English is not a primary language in Asia, they “started out with the hardest part of the puzzle” by building a language-agnostic algorithm that can be trained to analyse new languages, instead of having to use a translation tool that may be less accurate.
Ai Palette just announced that it raised US$4.4 million in its Series A funding in August, and aims to expand beyond Asia. They will be setting up offices in the United States and Europe these two years, and plan to launch more products to help clients with product innovation.
Sharing some interesting trends that Ai Palette had predicted so far, GanChoudhuri said, “As a result of the pandemic, people have been looking for food products to improve mental health, such as those that aid in better sleep. This is a market that is currently underserved, so we can expect to see more of such products in the next year or two.”
He also highlighted that peanut butter, a common ingredient in the breakfast and confectionery space, is now making waves in the alcoholic beverages section.
“Be it peanut butter whiskey or beer, you’ll be surprised to see how many of such products are in the United States market now!”