twig has been selected as one of the first companies to receive support through the UK’s Sovereign AI programme, unlocking access to national supercomputing infrastructure.
For Satnam, our CTO and Co-founder, this marks a key moment in scaling CANOPY – our family of multi-modal biological foundation models, and a core part of twig’s full-stack AI bioengineering platform.
“Biology isn’t a single prediction problem – it’s a system,” he explains. “To design something that works in production, you need models that can reason across that system – from DNA through to bioprocess and yield.”
This highlights a fundamental limitation in today’s BioAI landscape. Most models are built around single paradigms – protein structure prediction, DNA sequence modelling, or isolated optimisation tasks. While powerful, they don’t capture how biology behaves in reality, where outcomes emerge from interactions across multiple layers.
CANOPY is designed differently.
Built within bio:drive, our AI-native dry lab, CANOPY integrates biological data across sequences, pathways, phenotypes, and process conditions into a unified system-level model. Rather than optimising in isolation, it enables decisions that reflect real-world manufacturability from the outset.
Crucially, this is paired with grow:bot, our automated wet lab, forming a closed-loop learning system where every experiment generates structured, AI-ready data that feeds back into the model.
Until now, scaling this approach has been constrained by compute.
“Sovereign AI changes that,” Satnam says. “It allows us to move from validated models to truly scaled systems – integrating more data, training larger models, and pushing performance further.”
With access to national infrastructure, we’re now expanding CANOPY – increasing model scale, adding new biological modalities, developing benchmarks and deepening its integration into the experimental loop.