How McCormick and IBM Are Leveraging AI to Develop the Next Revolution in Spices

Creating the Next Flavor Sensation: How AI is Revolutionizing Food Product Development

Crafting the next iconic flavor is no easy feat. Few wake up thinking, "Today, I'm going to invent the next Oreo," and actually turn that thought into reality. Training to become a professional food product developer can take nearly two decades, not to mention the countless flavor iterations required to perfect a recipe that might rival favorites like Cool Ranch Doritos.

Thanks to IBM's Research AI for Product Composition, specifically its Philyra AI, spice manufacturer McCormick & Company is transforming flavor development. This collaboration leverages machine learning to reduce the flavor development timeline by up to 70%.

Last October, IBM introduced Philyra AI as a tool to accelerate the fragrance industry. "It uses advanced machine learning algorithms to analyze hundreds of thousands of formulas and raw materials," said Dr. Richard Goodwin, Principal Researcher at IBM's Computational Creativity Research Group. The AI identifies patterns and novel combinations, adapting formulas based on ingredient availability, user reactions, and dosage preferences.

Transitioning from fragrances to spices was seamless for IBM and McCormick. The Philyra AI’s initial success came from aiding Symrise in crafting new perfumes for O Boticário, a major beauty brand, set to launch later this year.

McCormick employs a 5-step Stage Gate system to create flavor profiles: ideation, development, testing, scale-up, and commercialization. Significant time is spent during the first two steps. For instance, when developing a new Tuscan Chicken recipe, researchers begin by mining the company database for previous recipes, then iteratively refine the flavor profile over several months and hundreds of trials.

IBM's AI significantly accelerates this testing process. By accessing McCormick's rich database of flavors and integrating sales and marketing data, the AI enables tailored recipes that cater to local tastes and dietary needs like Kosher, Halal, vegan, or low sodium, ensuring that alternatives stay true to the original flavor.

"This system helps us find flavors with broad appeal that could become the next iconic product," explained Dr. Hamed Faridi, McCormick's Chief Science Officer. The AI empowers McCormick's global R&D team of over 500 to pool their diverse expertise into a unified development effort, aiming to create the next standout product, be it a Philadelphia Cream Cheese or a Pringles chip.

Faridi likens the quest for an icon to searching for a needle in a haystack, but this AI is designed to simplify that search. Unlike human developers who might lean on familiar cheese flavors, the AI objectively evaluates all options in McCormick's database to identify the most appealing flavor.

Thus far, McCormick has utilized this innovative AI to develop three new Recipe Mix flavors: Tuscan Chicken, Bourbon Pork Tenderloin, and New Orleans Sausage, set for release by mid-year. The company plans to provide its entire R&D workforce access to this groundbreaking AI by 2021, paving the way for future culinary innovations.