[Disclosure: AFN’s parent company, AgFunder, is an investor in Ai Palette.]
Singapore-based startup Ai Palette is building out its capabilities to help food and beverage companies speed up their development process for new products with a new tool using generative AI to create new product concepts based on insights from its core trendspotting platform.
The new product concept generator—dubbed ‘Concept Genie’—can dramatically accelerate the innovation process by filling in the “missing piece” between Ai Palette’s core AI-powered insights platform (Foresight Engine) and its new concept screening tool (Screen Winner), claims the firm.
By trawling through images and online data in 16 languages from retail e-commerce platforms, menus, recipes, social media posts and search engines from 24 countries, Ai Palette’s Foresight Engine helps packaged food companies from Nestlé to Danone identify unmet consumer needs and emerging trends.
Until now, clients have been using this data to go back to their companies and come up with new product ideas that they can go back and screen using Ai Palette’s Screen Winner tool to determine if they have legs in a particular category or market, CEO Somsubhra GanChoudhuri told AFN.
The new feature, Concept Genie, sits between Foresight Engine and Screen Winner and uses generative AI to generates new product concepts that can immediately be screened on the platform, providing an end-to-end product innovation solution, he said.
“It completes the whole process from discovering an opportunity to generating a concept to screening it, making innovation possible at the click of a button.”
‘It speeds and de-risks the innovation process’
Human beings and more traditional tools are still needed to refine and validate concepts in the real world, stressed GanChoudhuri, who cofounded Ai Palette in 2018 with big data expert Himanshu Upreti, and has raised $5.5 million to date.
However, Ai Palette enables companies to identify trends and needs based on vast data sets in very specific target markets and generate tailored product concepts that meet these needs at the click of a button, without spending huge amounts of money, time, and energy using more traditional market research, said head of marketing, Americas, Michael Manes.
“It speeds and de-risks the innovation process. The fact that in one afternoon you can do some research, generate 20 product concepts, and test their market viability, takes away the fear of wasting time and running down trails that may not bear fruit.”
‘Concept Genie, build me a beverage concept with milk and lemon…’
Manes added: “Our AI historically was not generative, it was analyzing data, giving us results and making connections. Concept Genie takes prompts from our Foresight Engine about what people are looking for [based on recipes, social media posts, images, and search engine data] in a given market, and generates something new.
“Let’s say lemon and milk are trending. You say to Concept Genie, ‘Build me a beverage concept with milk and lemon’ that’s going to appeal to consumers [in a specific target market] and then you drop the concept into Screen Winner and it plots that for relevancy and originality. Is it differentiated vs what’s already on the market [based on Ai Palette’s analysis of data from menus, e-commerce sites and other online sources]?”
‘Now you can identify a market need, create, and screen a concept in a day’
GanChoudhuri explained: “We’ve been working on GPT [generative pre-trained transformer] technology for the last two, three, years. The generative AI [Concept Genie] piece is very new, but a few customers have been working with it and they’re excited that they can do everything in one go.”
GanChoudhuri, who previously worked at flavors & fragrances giant Givaudan, added: “Using more traditional approaches where you have to coordinate with multiple stakeholders and market research companies, even creating new concepts can take three to six months. Now you can identify a market need, create, and screen a concept in a day.”
Case studies: Cargill, Kellogg
Asked to provide concrete examples of how large CPG companies have used Ai Palette to date, he said: “One of the world’s largest alcoholic beverages companies used our platform to create new flavor variants of a vodka RTD drink and get them to market very quickly.
“In another case, one of Cargill’s customers wanted to enter the plant-based yogurt space in Indonesia. Using our platform, they could see it was a little premature to enter that particular market with this kind of product.”
Kellogg, meanwhile, used Ai Palette during the COVID-19 pandemic to scrutinize online content from Malaysia, the Philippines, Singapore and Thailand in four languages (Bahasa Melayu, Thai, Tagalog, and English) and identify trending recipes for foods incorporating breakfast cereal, such as crispy calamari.
It then used this data to launch a campaign on its social media channels featuring recipes incorporating its branded cereals.
Actionable insights
Given that new startups powered by AI are beginning to spring up all over the place promising to tell consumer goods companies what consumers want, how does Ai Palette stack up vs the competition in this space?
According to GanChoudhuri, being able to turn these insights into action by generating concepts and immediately screening them gives Ai Palette a significant advantage over startups just focusing on trends. Ai Palette is also unusual in its scope, incorporating data from multiple countries in multiple languages, and analyzing images as well as text, he claimed.
“We use computer vision to analyze images to get context about how a particular kind of food is being consumed indoors versus outdoors in a social setting, for example.”
Anticipating trends, rather than ‘constantly looking in the rear-view mirror’
Longer term, AI has the potential to transform new product development claimed GanChoudhuri. By looking at what people type into search engines about their health concerns, what they’re cooking at home, or what they’re researching online, companies can anticipate trends, rather than “constantly looking in the rear-view mirror,” he said.
“Companies need a predictive tool to identify not just the trends of today, but the trends of tomorrow.”
Building up a dataset enabling firms to analyze trends over time will also help companies get better at distinguishing between fads and trends, he predicted.
“Using big data analytics and AI we are able to identify and see how trends have evolved in the past and see which emerging trends, say, mimic the growth matcha had in its early days.”
