Rumen digital twin fine-tunes methane reduction strategies
A rumen digital twin can help unravel the complexity of rumen fermentation and guide the development of more effective methane-reduction strategies that support both lower emissions and improved cow productivity. We spoke with Aaron Schacht, CEO at BiomEdit, about this technology.
Livestock emissions are a significant source of greenhouse gas emissions, and a range of feed-based solutions aimed at reducing enteric methane is now moving beyond the research phase into early commercial adoption. Several feed additives on the market claim to deliver measurable methane reductions, in some cases reaching double-digit percentages. This represents meaningful progress and reflects substantial advances in scientific understanding and innovation.
However, an important question remains: are these products really effective if we aim for methane reduction and high animal performance at the same time? A number of commercially-available products indeed demonstrate clear methane-reducing potential, and some also show a positive correlation with productivity within specific production systems, whether in dairy or beef operations.
At the same time, notable differences exist between solutions. Certain additives may achieve strong methane reductions but offer limited benefits for animal performance, while others support productivity yet have a more modest impact on emissions.
“Ideally, a methane-mitigation strategy should deliver both outcomes: improved productivity and reduced emissions. There is no inherent reason these objectives must be in conflict, yet in practice they are often treated as trade-offs. The key challenge, therefore, is how to unlock solutions that achieve both simultaneously. Using AI, and more specific, digital twins, helps us do that,” said Aaron Schacht, CEO of BiomEdit, in his presentation at the Animal Health, Nutrition & Technology Innovation event (AHNTI), held 2-4 March 2026 in London, UK, and organised by Kisaco Research.
Explainable AI for livestock
At AHNTI, many talks delved into how artificial intelligence (AI) can be applied to solve animal health challenges or to improve diagnostics (outcomes and speed). Also for innovation company BiomEdit, carved out of Elanco Animal Health 4 years ago, AI is an important part of its research. The company focuses on microbiome discovery to better understand microbial ecosystems, and on synthetic biology to engineer beneficial probiotic bacteria. In addition, explainable AI is a major pillar in the company’s life sciences research and is now being applied to address animal health issues but also climate-related challenges, including methane emissions from livestock production.
Schacht: “Explainable AI focuses on developing predictive models capable of rapidly analysing large volumes of data while also making the relationships between variables transparent, thereby improving our understanding of complex biological systems. In livestock research, a clear example of such a complex system is the cow’s rumen and the process of methanogenesis (the production of methane by the rumen microbiome) occurring within it. When additional factors such as genetics, the composition of the rumen microbiome, and the animal’s feed intake are taken into account, the system becomes even more complex. The methane challenge in the livestock industry is a perfect example of how AI can be used to address complex climate challenges.”
Building a rumen digital twin
According to Schacht, AI can support our efforts to better understand the complexity of the cow as a biological system. “By improving this understanding, we can more accurately predict methane production and determine how methane mitigation strategies influence key production parameters,” he said. “With these insights, we can design next-generation methane-reducing feed solutions that are more effective overall. As noted earlier, many feed additives currently on the market are capable of significantly lowering methane emissions but have limited impact on productivity, while others support productivity without substantially reducing methane. The objective, therefore, is to use deeper biological insights to develop solutions that successfully achieve both outcomes.”
The methane challenge has driven BiomEdit to expand both its knowledge base and its technological capabilities in this field. Schacht explains: “In 2023, we received a US$4.5 million grant from the Gates Foundation for our ruminant methane programme. Last year, we were awarded an additional US$2 million from the Bezos Earth Fund, which supports the use of AI to develop climate solutions across multiple industries, and we were the only livestock-focused company selected. This funding enabled us to begin developing a rumen digital twin, an AI-based model designed to predict the relationship between inputs and outputs in dairy and beef production, with respect to environmental sustainability and productivity. The model uses large integrated datasets that combine publicly-available and proprietary data on genetics, caloric and nutritional intake and microbiome composition.”
Model helps fill knowledge gaps
The first-generation rumen digital twin developed by BiomEdit is now ready and used in research and data-gathering. Although the model was initially built using dairy cow data, it can accept inputs from other ruminant species. Trial runs using sheep data (in this case from Australia) produced reasonable predictions for both methane emissions and production performance. These results indicate that the AI model generalises across species, geographic regions, and differences in microbiome composition. The system is a generative foundation model designed to capture the context-dependent relationships between ruminants and their rumen microbiomes.
The synthetic data generated by the first generation rumen digital twin now begins to mine data to refine BiomEdit’s efforts to fine-tune the model and fill existing knowledge gaps. This capability is important because global data on ruminant emissions, performance, genetics, and interventions do exist, but they are fragmented across studies and stored in inconsistent formats. As Schacht explains: “Not every research group measures the same variables, or measures them in the same context, but there is an underlying, highly complex biology that connects them. Our digital twin model helps unravel this biological complexity and provides insight into why certain methane-mitigation technologies produce inconsistent results. It also allows us to explore future scenarios. For example, if we define the methane emission level and productivity we want to achieve, the model can help design the corresponding microbiome profile, together with the feeding strategy required to increase productivity while reducing methane emissions.”
Mitigation efficacy increased by 70%
Given the strong focus on methane reduction, for clear climate reasons, but also because methane represents a loss of protein and energy for the animal, we need better tools to truly understand and manage this system. As the microbiome in the rumen is key for the formation of methane, the key question is: If we know the microbiome of the cow, can we predict its performance and its methane emission? If so, what we can we do with this information relative to advancing solutions?
The rumen digital twin, developed by BiomEdit, replaces lengthy and expensive animal experiments to discover methane mitigation technologies and tailor them to the conditions where they are most effective. “With this information, we can tailor ruminant diets to reduce methane and improve productivity and even design microbiomes for desirable phenotypes. I strongly believe that by knowing what works, and where, in the complex rumen environment we can both fine-tune and increase the efficacy of methane mitigation technologies by 70%,” Schacht concludes.
Join 13,000+ subscribers
Subscribe to our newsletter to stay updated about all the need-to-know content in the dairy sector, two times a week.
