Hiverge, a start-up launched by former Google DeepMind research leaders, has secured £3.7m in seed funding for its algorithm factory, aiming to solve a hidden yet critical business challenge: discovering and deploying algorithms that help organisations rapidly improve their operations at scale. Flying Fish Ventures led the round, with Ahren Innovation Capital and Jeff Dean among the other investors.
Algorithms do not just dictate what appears in search results or social media feeds; businesses across industries, from retail to cloud infrastructure, rely on them for a huge number of critical operations. Yet current algorithms fall short of today’s requirements. Systems are growing in complexity with businesses needing to make accurate decisions in milliseconds, while existing computational resources remain constrained. Designing tailored algorithms to meet specific use cases is both expensive and requires rarely available expertise, forcing most companies to rely on one-size-fits-all solutions. The result is that businesses are rapidly approaching the limits of what traditional, human-led approaches can deliver, curtailing growth and putting efficiency gains out of reach.
Hiverge aims to solve these problems by giving businesses access to tailored and efficient algorithms that continually adapt to changing environments. Unlike traditional software engineering and trends such as vibe coding, which build front-end applications and websites, Hiverge’s technology uses program synthesis to create an algorithm factory: automatically writing and optimising code tailored specifically to each user’s unique challenge at scale.
This AI discovery engine is at the forefront of a new category of algorithmic discovery that originated within Google DeepMind. Hiverge’s founders led the team that spearheaded this groundbreaking category with AlphaTensor and Fun Search. They also worked on the early stages of Google’s AlphaEvolve agent, a project that improved the efficiency of Google’s data centres, chip design and AI training processes.
Hiverge’s platform, the Hive, has multiple uses, from accelerating AI training to helping businesses plan how they allocate resources. For example, the Hive discovered fast and resource-efficient training algorithms improving over existing human-engineered approaches by more than 15% on well-established AI benchmarks. Additionally, on an industrial planning problem posed by Airbus, the Hive discovered sophisticated long-term planning algorithms that run several orders of magnitude faster than existing classical planners.
Hiverge’s founders include CEO Alhussein Fawzi, a former Google DeepMind research scientist and MIT Innovators Under 35 award winner, CTO Bernardino Romera-Paredes, also a former Google DeepMind research scientist and part of the Nobel prize-winning AlphaFold team, and chief science officer Hamza Fawzi, a professor in the Department of Applied Mathematics and Theoretical Physics at the University of Cambridge and a specialist in mathematical optimisation.
