The global cocoa market is facing unprecedented supply chain volatility, climate-driven shortages, and soaring costs. As a result, the food industry is urgently seeking sustainable ways to recreate the rich, comforting experience of chocolate—without relying on conventional cocoa beans.
While several innovative startups are tackling this challenge using different feedstocks—such as sunflower seeds (Planet A Foods) or fava beans (Nukoko)—at Mycosortia, we are taking a deep-tech approach. Using our proprietary, AI-powered biotransformation platform, AuroGenesis™, we upcycle agricultural side-streams like soy okara, alongside other readily available staple grains and legumes, into authentic chocolate flavor profiles.
The Science of Flavor: Recreating the Maillard Reaction
To understand how bean-free chocolate works, you first have to look at traditional cocoa processing. The signature flavor of real chocolate is generated during roasting through the Maillard reaction—a chemical reaction between amino acids and reducing sugars produced during the natural fermentation of cocoa beans.
Without cocoa, you can still trigger the Maillard reaction using amino acids and sugars derived from alternative fermented feedstocks. However, replicating the exact sensory complexity and depth of real cocoa has historically been a massive challenge.
This is where artificial intelligence changes the game.
How AuroGenesis™ Uses Machine Learning to Replicate Cocoa
Recreating the exact amino acid and sugar precursor balance—and matching the precise reaction conditions needed to emulate cocoa—requires navigating millions of variables. Machine learning makes this level of precision possible.
Here is how our AI-powered process works:
- Sustainable Feedstocks: We start with around 10 accessible ingredients, featuring upcycled soy okara as our core base alongside other staple grains and beans.
- Mixed-Culture Solid-State Fermentation: The feedstocks undergo controlled biotransformation using proprietary microbial co-cultures.
- AI Optimization: Our machine learning models analyze and optimize three critical variables simultaneously:
- Feedstock blend compositions
- Fermentation parameters (temperature, time, microbial balance)
- Thermal processing and roasting conditions
- Data-Driven Precision: Trained on thousands of sensory data points, the model predicts the exact formulation and process parameters required to achieve the ideal chocolate profile.
The Proof is in the Tasting: Human Sensory Validation
Does AI-predicted flavor actually taste better to real humans? We put it to the test in a sensory trial with 60 panelists, comparing two chocolate alternatives: one developed using traditional trial-and-error formulation, and one generated using our machine learning predictions.
The results were clear:
- Higher Overall Acceptance: While panelists liked both versions, the AI-optimized chocolate alternative achieved a significantly higher overall acceptance score.
- A Superior Profile: The panel overwhelmingly agreed that the product created via AI prediction was distinctly “more chocolatey” in flavor and complexity.
What did the tasters have to say? Feedback from our 60 panelists confirmed both the authenticity and versatility of the prototypes. Participants praised the alternatives for having a “good chocolate flavour, sweet enough and tastes like a regular choc bar” or feeling more like “a commercial chocolate.” While distinct profiles emerged—with Sample 465 leaning closer to traditional cocoa and Sample 132 offering a pleasant maltiness—the overwhelming consensus was that “both are nice!” and successfully hit the mark.
By allowing machine learning to direct the precursor synthesis, we achieved a level of sensory fidelity that traditional food formulation methods simply couldn’t match on their own.
Beyond Chocolate: A Universal Platform for Novel Natural Flavors
The beauty of the AuroGenesis™ platform is its versatility. The platform isn’t limited to cocoa; machine learning gives us the flexibility and precision to tune amino acid and sugar compositions to generate a wide variety of clean-label, natural flavor matrices.
Whether it’s developing savory umami and kokumi modifiers or cheesy flavor profiles, our AI-driven biotransformation allows us to deliver tailored, flexible flavor solutions for food manufacturers worldwide.
As we continue to refine our models and scale production, Mycosortia is excited to lead the charge toward a more sustainable, resilient, and delicious future for the global food industry.
To learn more about our AuroGenesis™ platform or partner with us on clean-label ingredient solutions, get in touch at contact@mycosortia.com.