Research project
AI Platform for Sustainable Protein from Agri-Food Waste
- Start date: 1 January 2026
- End date: 31 December 2027
- Funder: Bezos Earth Fund
- Value: $1,998,000
- Partners and collaborators: CSIRO (Commonwealth Scientific and Industrial Research Organisation, Australian Government)
- Primary investigator: Professor Nicholas Watson
- Co-investigators: Célia Ferreira, Dr. Darren Greetham, Alex Bowler
- External co-investigators: Kai Knoerzer, Jordan Pennells, Carol Hartley, Stephen Wan, Necva Bolucu, Pablo Juliano (CSIRO )
Nearly 29% of the world’s population faces food insecurity, yet millions of tonnes of nutrient-rich agri-food waste is generated globally each day. This waste can be upcycled into sustainable, high-quality microbial protein (e.g. yeast, fungi, bacteria and algae) via fermentation for use as ingredients in existing foods or to produce novel meat and dairy alternatives. However, microbial fermentation processes are difficult to design, optimise and scale up, especially for complex, perishable and highly variable waste streams.
This limits their investment attractiveness and economic viability, especially for small and medium enterprises (SMEs), and ultimately their global adoption. Our open-access AI platform addresses these challenges by streamlining the development of microbial protein production processes. Leveraging advances in modern AI, the platform consists of a Waste to Protein AI Engine coupled with an adaptive user interface.
The platform will enable global users (agri-food waste producers, fermentation or bioprocess companies, researchers, and food producers) to input their waste composition, volume, and location and immediately receive actionable insights, tailored to their needs, such as optimised fermentation parameters, microbial strain recommendations, projected protein yields and economic returns.
Our AI platform will also provide users with an R&D roadmap that recommends critical trials to perform during process development and scale-up to reduce experimentation needs by up to 90%. Finally, it will provide partnership recommendations for areas where they lack expertise, facilities or market access through a dedicated user database. Our proposed AI solution has been co-created with over 40 diverse food system stakeholders, including future beta users, leveraging established partnerships in lower to higher-income countries.
The utility of our AI solution will be grounded in three regional case studies (the UK, Australia and Indonesia) where it will be used to produce tasty, nutritious and affordable upcycled protein products in collaboration with local partners. The AI solution and developed food products will be showcased at 11 engagement workshops across six continents, attracting new users and driving global adoption.
Through thoughtful design and strategic global partnerships, our AI solution delivers transformative environmental and societal benefits by democratising sustainable protein production to reduce agri-food waste and food insecurity. It does not just model pathways — it redefines who can innovate, where protein can be made, and how fast sustainable change can happen.