Food System Innovations (FSI) has been awarded a Phase I grant from the Bezos Earth Fund’s AI Grand Challenge for Climate and Nature. With this $50,000 award, FSI will collaborate with Stanford PhD candidate Anna Thomas to create an open-source AI assistant aimed at revolutionizing how food scientists formulate sustainable proteins.

The challenge, launched with up to $100 million in funding, is designed to unlock bold, AI-powered ideas that can directly address the climate and biodiversity crises. FSI’s selection signals growing recognition that food system innovation—particularly in plant-based and alternative proteins—is an essential component in building a more sustainable planet.

“Our recent results, accepted to the International Conference on Machine Learning this year, suggest that LLMs, combined with other algorithms and datasets, can support a sustainable protein transition,” said Thomas. “In particular, our results indicate that LLMs can be useful for revising formulations in response to sensory panel feedback.”

Building Smarter Food Systems, Faster

FSI’s work leverages the potential of Large Language Models (LLMs) to significantly reduce the time required for food scientists to iterate and improve new product formulations. This could dramatically accelerate the market availability of better-tasting, more affordable, and more sustainable food options.

According to Thomas’ research, LLM-supported teams spent 45% less time on average adjusting formulations—compared to 22% when working with human experts alone. With this edge, AI could help companies respond more efficiently to consumer sensory feedback, improving taste and texture at record speeds.

“The protein transition will be driven by taste, price and convenience,” said Caroline Cotto, Director of NECTAR, an initiative of FSI. “Our goal with NECTAR is not just to hand brands sensory data but to actually empower them with a path forward.”

According to Cotto, NECTAR intends to utilize this partnership with Thomas to improve the implementation of their research into the R&D of food manufacturers. The organization conducts large scale blind taste tests with flexitarian eaters to determine which plant-based products are competing on a taste level with their animal-based counterpart. Maximizing efficiency of R&D will lead to better tasting products, greater consumer adoption and repeat customers for retail and foodservice operators.

What’s Next for Food and AI?

FSI, NECTAR and the Bezos Earth Fund are inspiring food innovators across the world to tap into the power of AI to improve the way food is consumed. As we open our minds to the possibilities, here are five creative ways continued AI advancement could improve the global food system:

1. Sensory-Driven Product Iteration

LLMs can help R&D teams synthesize real-time consumer panel feedback into immediate formulation suggestions—saving months of manual analysis and allowing brands to rapidly iterate toward consumer-preferred products.

2. Localized Ingredient Optimization

AI models could analyze regional supply chains and dietary preferences to recommend ingredient swaps that maintain flavor profiles while lowering carbon footprints—e.g., swapping imported pea protein with local pulses.

3. Precision Nutrition for Institutional Foodservice

School lunch programs, hospitals, and military menus could benefit from AI-generated recipes that meet strict dietary guidelines while using plant-based ingredients that appeal to mainstream palates.

4. Recipe Development for Low-Infrastructure Environments

AI could help humanitarian aid groups and emerging markets develop high-protein, shelf-stable recipes using limited equipment or ingredients—supporting nutrition resilience under climate stress.

5. Transparent Labeling & Compliance

AI can auto-generate nutrition labels, allergen statements, and sustainability impact reports in multiple languages, making it easier for small brands to enter global markets with full compliance and clarity.

Looking ahead, FSI is now eligible for a $2 million Phase II implementation grant from the Bezos Earth Fund. If selected, their work could scale dramatically—offering open-source tools that make alternative protein innovation faster, more affordable, and more accessible for food manufacturers around the world.

In an industry where taste is king and time is money, this AI-forward approach has the potential to rewire the way the world eats—making delicious, sustainable food the default, not the exception.

Learn more about the program at fsi.org and nectar.org.