Saffiera Karleen is Business Development Manager at Corbion, an Amsterdam-based specialty ingredients company with more than a century of expertise in fermentation science, food preservation, and shelf-life extension. Corbion supplies food manufacturers across the globe with fermentation-derived ingredients, lactic acid and its derivatives, functional blends, and algae-based nutritional solutions.
In this guest post, Karleen draws on her background as a trained food formulator to explore how artificial intelligence is reshaping product development in the plant-based sector, and what that means for safety, shelf life, and sustainable innovation.
The plant-based food sector has evolved rapidly. Early products often required compromises in taste, texture, or shelf life, but today’s consumers expect plant-based foods to perform just as well as conventional alternatives. Safety, freshness, and appealing sensory profiles are now prerequisite requirements. For food manufacturers, the challenge is delivering on those expectations consistently and at scale, while maintaining sustainable supply chains within a tight development timeline.
Product development in plant-based systems involves complex ingredient interactions, where minimal formulation adjustments can significantly influence microbial stability, texture, and overall product performance. Artificial intelligence is increasingly used by food scientist to navigate this complexity, enabling faster analysis of large datasets and modelling product behavior under different conditions. Rather than replacing scientific expertise, AI is supporting the food innovator by simplifying the translation of complex data into practical and confident formulation decisions.
From trial-and-error to predictive development
Historically, food formulation has relied heavily on empirical experimentation. Scientists will spend tremendous amount of time in observing the impact while adjusting variables such as pH, water activity, ingredient combinations, and preservation strategies until the right balance between safety, shelf life, and sensory performance is achieved.
This process becomes particularly demanding in plant-based products. Without animal-derived proteins and fats, plant matrices behave differently, influencing microbial growth patterns, moisture retention, and product stability. Formulation adjustments that appear minor can have significant downstream effects.
AI-supported modelling tools are helping shift product development from trial-and-error toward predictive analysis. By processing large datasets drawn from microbiological studies, laboratory screenings, and real-time product testing, these systems can simulate how different formulations may behave under specific storage and processing conditions. Shelf-life trajectories, microbial growth risks, and ingredient interactions can be assessed digitally, allowing research teams to focus their validation work on the most promising formulations.
The result is a more efficient development process, where scientific insights can be applied earlier and more strategically.
Data-driven insights for safer plant-based foods
Predictive modelling is particularly valuable where manufacturers must balance safety, shelf life, and ingredient transparency to meet consumer demand.
Fermentation-derived preservation ingredients, such as cultured vinegars and organic acids, play an important role in controlling microbial growth, ensuring safety and extending freshness. Their effectiveness, however, depends on precise interactions between formulation parameters including pH, salt concentration, water activity, and storage temperature. Digital modelling tools help formulators explore these relationships more systematically, before committing to physical testing cycles.
A concrete example is the Corbion Listeria Control Model (CLCM), a predictive microbiology tool built on more than two decades of microbiological challenge data and enhanced through AI-supported analysis. The model helps estimate how Listeria monocytogenes may behave under different formulation and storage conditions, integrating data from conducted challenge studies, laboratory screenings, and peer-reviewed research. For plant-based manufacturers working with novel ingredient systems, this type of insight can significantly accelerate product development while strengthening confidence in shelf-life performance.
The human element remains central
Despite the growing role of AI in food development, human expertise remains essential.
Food is not purely a technical system. It is also a sensory experience shaped by taste, texture, and consumer expectations that no algorithm can fully replicate. Trained sensory panels, experienced application scientists, and formulation specialists provide a depth of contextual understanding that sits beyond the reach of predictive models alone.
The most effective innovation models combine computational analysis with human expertise. The CLCM illustrates this well: AI enhances the model’s predictive accuracy, but it is built on and validated against decades of real-world scientific data. The model informs; experienced application scientists interpret and translate.
At Corbion, this means bringing together digital modelling tools with deep expertise in fermentation science and food preservation. Data-driven insights accelerate early formulation work, while application scientists ensure that final products meet safety, performance, and sensory requirements. AI acts as a tool that strengthens scientific decision-making, not one that replaces it.
Sustainability through smarter formulation
Beyond accelerating innovation, data-driven modelling supports broader sustainability goals.
Food waste and recalls due to safety concerns remain one of the most significant challenges within the global food system, and shelf-life performance sits at the heart of it. Products that spoil prematurely represent wasted resources throughout the supply chain, from raw materials and processing energy to transportation and retail.
AI-enabled modelling helps manufacturers design products with more reliable shelf-life performance by identifying potential stability risks earlier in development. Development teams can reduce failed trials, optimize ingredient use, and design preservation strategies that support longer and more consistent freshness.
Improved predictive modelling also strengthens food safety management and reduces the likelihood of costly product recalls. Beyond the immediate financial impact, a recall carries longer-lasting consequences: damaged consumer trust and eroded brand credibility that can take years to rebuild. For plant-based brands whose market position rests on a promise of quality and transparency, that reputational risk is not abstract. Confidence that a product will perform as intended, from production through to the end of its shelf life, is both a safety outcome and a brand protection strategy.
Looking ahead
The plant-based sector is entering a more demanding phase of maturity. The early enthusiasm that drove initial market growth has evolved: consumers are more discerning, competition has intensified, and regulatory expectations continue to tighten. Buying plant-based is no longer a statement in itself; the product has to genuinely deliver on taste, safety, and shelf-life stability.
AI-supported modelling tools are helping manufacturers navigate this complexity by transforming large volumes of scientific data into actionable insights. When combined with human expertise and deep ingredient knowledge, these technologies allow food developers to move from reactive troubleshooting toward more predictive, confident innovation.
For the plant-based food sector, that shift represents an important step forward: enabling the development of products that are safer, fresher, and more sustainable, while meeting the expectations of both consumers and regulators.
That is a future worth formulating for.
