AI expert Umami unveils 'virtual marine cell' to speed up research

Marine biotechnology company Umami Bioworks has launched what it says is the world’s first “virtual marine cell”, an AI-driven model that it says can dramatically speed up aquaculture research and development.

Singapore-based Umami says the constraint-based metabolic model simulates how aquatic cells grow, respond to nutrients and stress, and produce key compounds. It simulates the inner workings of aquatic species with unprecedented precision, enabling faster R&D, sharper predictions, and radically reduced experimentation for aquaculture, cultivated seafood, and marine bioactives.

Using species-specific datasets from tuna, salmon, eel, and other high-value organisms, it models how real cells grow, respond to nutrients, react to stress, produce critical compounds, and transport them out of the cell. By learning directly from biological and environmental data, the platform replaces years of slow trial-and-error with instant predictive insights, says Umami.

“Aquaculture’s biggest bottlenecks like growth variability, disease, stress sensitivity, feed inefficiency, and unstable yields all stem from cellular behaviour that has been historically difficult to measure,” says Umami in a press release.

“The Virtual Marine Cell enables researchers and producers to predict growth rates and optimise culture conditions, simulate disease and immune responses, improve feed and nutrient strategies, model stress tolerance and temperature sensitivity, and identify genetic and metabolic markers tied to performance, stress, or nutritional content. This means faster breeding programmes, more resilient stock, and sharply reduced biological risk.”

The AI model can also enable scientists seeking marine lipids, peptides, enzymes, and bioactives for the skincare, nutrition, pharmaceuticals, and wellness markets to achieve results “in days instead of years”, says Umami.

Ashwath Bendre, product manager at Umami Bioworks, said: “Marine biology has never had a computational engine like this. The Virtual Marine Cell cuts through biological uncertainty and lets the industry move with software speed instead of lab-bench speed.”

The company says it is already collaborating with leading global seafood, aquaculture, biotech, and consumer-goods companies to apply the virtual marine cell to real-world challenges across production performance, resilience, and bioactive discovery.

Cell culture expertise

In March, Umami Bioworks announced that it was using its datasets and artificial intelligence (AI) expertise to expand beyond research into cell-cultivated seafood and into traditional aquaculture, wild fisheries, and seafood supply chains.

It announced the latest iteration of ALKEMYST, a domain-specific AI foundation model originally developed to accelerate cell-cultivated seafood production and improved to support the broader seafood value chain.

The company said the AI model has been trained on a comprehensive dataset encompassing public and proprietary marine genomics, metabolomics, and seafood sensory data, plus a rapidly expanding proprietary dataset that includes production process parameters and organoleptic attributes specific to seafood species.

Data-driven decisions

“This next-generation version of ALKEMYST is a huge step forward in AI-driven seafood innovation,” said Umami's co-founder and chief executive Mihir Pershad.

“We’ve expanded ALKEMYST beyond cultivated seafood, enabling businesses across the entire seafood value chain to make data-driven decisions that improve efficiency, sustainability, and resilience. This is more than just an AI model - it is a transformative tool for the future of global seafood.”

Umami Bioworks’ investors include Netherlands-based aquaculture investor Aqua-Spark, and Japanese tuna and trout farmer Maruha Nichiro.