What Confluence Labs does
Confluence Labs is an AI research lab focused on learning efficiency, meaning how quickly a model can learn from limited data or few experiments. It builds AI models that learn from experience, aiming to perform well in domains where data is sparse or costly to obtain, such as designing molecules, discovering physics, and engineering materials.
Key capabilities
- Research on learning efficiency and data-efficient AI
- Methods that direct language models to write code describing the transformation in a reasoning task
- AI designed to plan effective experiments in data-sparse domains
Who it's for
Confluence Labs targets scientific and research domains where collecting data is expensive and few-shot reasoning matters. Its differentiation is a focus on maximizing learning from scarce data; the lab reported state-of-the-art performance on the ARC-AGI-2 benchmark, scoring 97.9% on the public evaluation at roughly $12 per task. The company is part of Y Combinator's W26 batch.