Flower Raises $3.6M to Expand Federated Learning Platform for AI
Flower, a startup co-founded by Daniel Beutel from the University of Cambridge, has secured $3.6 million in pre-seed funding. The company addresses a crucial limitation in AI research, focusing on decentralized data for training models. Beutel highlights the drawbacks of relying solely on public data for AI training. He points out that while public data has its uses. It’s limited in scope, whereas distributed data, trapped on devices or within organizational silos, holds much more potential.
Launched in 2020, Flower’s platform empowers developers to train AI models using data from various devices and locations. Recently, they introduced FedGPT, which allows training large language models (LLMs) using sensitive internal data without compromising privacy. This is especially relevant as AI faces increasing regulations and data privacy concerns.
Flower is collaborating with Brave, an open-source web browser, on Project Dandelion. Which aims to build an open-source federated learning system for Brave’s user base of over 50 million. Flower’s developer community has grown to over 2,300, including prominent Fortune 500 companies and esteemed academic institutions like Porsche, Bosch, Stanford, Oxford, MIT, and Harvard.
Investors, including First Spark Ventures, Hugging Face CEO Clem Delangue, and others, support Flower’s vision. Beutel emphasizes the importance of collaboration in providing a comprehensive range of open-source federated techniques for the AI community.
In conclusion, Flower’s successful $3.6 million funding round, led by Daniel Beutel, is set to drive the expansion of its platform focused on decentralized AI training. With the launch of FedGPT and the partnership with Brave, Flower is positioned to play a pivotal role in revolutionizing AI training methodologies while ensuring data privacy and compliance.
