Liquid AI, a startup building a new class of adaptable artificial intelligence models, has raised $37.6 million in seed funding to bring its technology to market. The round, led by top VC firms OSS Capital and PagsGroup, values Liquid AI at $303 million pre-money.
Liquid AI was founded earlier this year by Daniela Rus, director of MIT’s Computer Science and Artificial Intelligence Lab, and three other researchers. The company aims to commercialize the team’s groundbreaking research on liquid neural networks.
Unlike traditional AI models comprised of static neural connections, liquid neural networks can modify their own architecture to better handle new data and situations. This adaptability promises to make them more reliable for real-world deployment.
Better Reliability, Efficiency with Less Data
A key benefit of liquid networks is robustness: by continuously tuning their structure to current inputs, they can sustain reliable performance amid changing real-world conditions. For example, an autonomous vehicle powered by liquid AI may handle rain or snow more gracefully without additional training.
These AI systems also require far fewer computational resources - neurons and tuning parameters - to operate. This allows them to run efficiently using less infrastructure while maintaining accuracy.
Moreover, with fewer moving pieces to orchestrate decisions, researchers can more easily analyze model reasoning after the fact. Such transparency helps ensure that AI-powered systems behave safely and ethically.
Positioned to Lead Next Wave of AI
The expert backers and nine-figure valuation cement high expectations for Liquid AI in shaping the future of artificial intelligence.
Joining lead VCs in the round are Samsung Next, WordPress developer Automattic, and Red Hat co-founder Bob Young, signaling a breadth of commercial possibilities.
The startup plans to allocate funds toward developing versatile foundation models as well as a full-stack platform for customers to build specialized liquid networks tailored to any field.
With adaptable infrastructure powering automated systems across sectors like autonomous transport and personalized medicine, Liquid AI’s novel approach could enable the next leap in reliable, efficient AI.