INNOVATION, SCIENCE & TECHNOLOGY

Artificial intelligence is beginning to move beyond optimizing fermentation conditions. Researchers are exploring whether it can help design the microbial communities that determine flavour, texture, functionality and consistency before fermentation even begins.

For centuries, food fermentation has depended on microorganisms doing extraordinarily complex chemistry, often with humans understanding only part of what happens inside the system. Modern manufacturers have brought far greater control to the process, yet developing starter cultures can still involve extensive screening, experimentation and trial and error.

A new scientific review suggests that could be changing. Researchers writing in Trends in Food Science & Technology describe an emerging approach that combines artificial intelligence, machine learning, multi-omics and metabolic modelling to move starter-culture development from empirical screening toward more rational, data-driven design.

The implication is significant. The next frontier of AI in food manufacturing may not simply be machines that inspect products or algorithms that optimize production lines. It could involve using computational tools to predict what microorganisms will do before they are ever deployed in fermentation.

From Finding Microbes to Designing Fermentation

Starter cultures influence some of the characteristics consumers care about most: flavour, aroma, texture, nutritional functionality, stability and, in some applications, safety. Traditionally, identifying the right microorganisms and combinations has required extensive experimental screening. The emerging model is different.

By combining genomic, transcriptomic, proteomic and metabolomic information with machine learning, researchers can attempt to identify relationships that would be difficult to resolve through conventional analysis alone. The recent review highlights applications ranging from predicting microbial interactions and metabolic functions to modelling flavour-related pathways. Genome-scale metabolic models and emerging digital-twin approaches could further help researchers simulate how microorganisms behave under different conditions.

Separate research published earlier this year similarly described machine learning as an emerging tool for constructing microbial communities for food fermentation, while another review examined combining metagenomics, multi-omics and AI for microbiome engineering and predictive food bioprocessing. The direction is increasingly clear: fermentation research is moving from asking which strain works toward which microbial system should be designed for a particular outcome.

What Could This Mean for Manufacturers?

Consider a manufacturer developing a fermented plant-based product. The challenge is not simply achieving fermentation. The manufacturer may simultaneously want to reduce undesirable flavour compounds, generate desirable aromas, improve texture, increase nutritional accessibility and maintain consistent performance during commercial production.

Recent research on plant-based fermentation emphasizes that these outcomes depend strongly on the interaction between substrate, microbial strain and processing conditions, with AI-guided optimization now being investigated alongside controlled starter cultures and omics technologies to improve reproducibility and product-specific performance. Instead of testing combinations almost entirely experimentally, future development systems could increasingly use computational models to narrow the possibilities first, potentially reducing development cycles and letting R&D teams target microbial combinations toward specific product characteristics.

Dairy provides another example. A review published August 19 examined the integration of multi-omics and AI in modelling dairy fermentation, reflecting growing scientific interest in using richer biological datasets to understand and ultimately control complex fermentation systems.

The Bigger Opportunity Is Precision

This should not be confused with precision fermentation alone, which generally uses microorganisms as production platforms for targeted molecules such as proteins, enzymes or fats. AI-designed food fermentation could have a broader application: understanding and influencing entire microbial ecosystems responsible for transforming food.

That creates the possibility of increasingly customized starter cultures designed around outcomes such as flavour formation, texture development, fermentation stability or functional metabolite production. It could eventually make microorganisms themselves a more programmable component of food product development.

But AI Has Not Solved Fermentation

The technology remains far from a push-button system where a manufacturer enters a desired flavour profile and an algorithm produces the perfect starter culture. The researchers identify substantial obstacles. Fermentation datasets are often fragmented and generated under different experimental conditions. Biological systems are highly complex. AI models can be difficult to interpret, and predictions that work computationally or at laboratory scale still need validation under real manufacturing conditions.

There are also practical questions around data standardization, industrial integration, cost, workforce capabilities and regulation. A broader 2026 review of AI-driven food bioprocessing similarly identifies data quality, model interpretability and regulatory issues among the barriers separating computational potential from industrial implementation. AI is not replacing fermentation science. It is becoming another tool through which fermentation scientists interrogate increasingly complex biological systems.

What GFI Is Watching

The most important development may ultimately be the convergence of technologies: AI, multi-omics, metabolic modelling, synthetic biology and fermentation engineering, each progressing independently but together capable of creating a new generation of food-biotechnology platforms that move from microbial discovery toward microbial system design.

For food companies, ingredient suppliers and biotechnology startups, this raises a strategic question. If fermentation becomes increasingly predictable and programmable, will competitive advantage come from owning the microorganism, the manufacturing process, or the datasets and models that explain how the microorganism behaves? The answer may determine where some of the most valuable intellectual property in future food fermentation resides.

The next food innovation race may therefore be happening somewhere manufacturers have always depended on, but are only beginning to understand computationally: inside the microbe.

---

Sources

1. Shi, H., Zhang, C., Zhao, J., Zhai, Q., Chen, W. & Tian, F. AI-driven design of next-generation food starter cultures: From empirical screening to intelligent customization. Trends in Food Science & Technology, 174, 105835, 2026.
2. Trends in Food Science & Technology. Machine learning-driven microbial community synthesis: A new paradigm for intelligent upgrading of food fermentation. 2026.
3. Trends in Food Science & Technology. AI-driven food bioprocessing: How far its integration with practice? 172, 105662, 2026.
4. Terzioğlu, M.E. & Tekgül, Z.Ç. A Review on Modeling the Fermentation Process of Dairy Products Using Multi-Omics and Artificial Intelligence Approaches. Fermentation, 12(8), 391, August 19, 2026.
5. Trends in Food Science & Technology. Food metagenomics integrated with multi-Omics and AI: Functional insights, microbiome engineering and predictive bioprocessing for sustainable food preservation. 169, 105566, 2026.
6. Sam, F.E. & Aryee, A.N.A. Fermented plant-based foods: Insights and emerging trends. Food Research International, 2026.