Symbolica

London-based Symbolica is a research lab pursuing a contrarian path in AI: neuro-symbolic architectures grounded in category theory, aimed at building models that reason in structured, interpretable ways rather than relying purely on scale. The thesis is that mathematically principled structure can yield more reliable and efficient AI than ever-larger black-box networks. It has released open-source agent frameworks alongside its research. It appeals to those who think the current LLM paradigm will need fundamentally new foundations.