Flourish AI is a newly established neuro-AI company that aims to solve two of the biggest limitations of modern artificial intelligence: high energy consumption and the inability of current AI systems to learn continuously after training. Founded by neuroscientist and entrepreneur Reardon, former Amazon executive Rob Williams, and a team of leading researchers, the company seeks to create AI systems that operate more like the human brain. Their vision is to develop a synthetic intelligence that matches the learning efficiency, adaptability, and low power requirements of biological intelligence while avoiding the enormous computational demands of today’s large language models (LLMs).
The idea attracted strong investor interest. After reviewing a proposal describing the company’s mission, Jeff Bezos invested $50 million and later increased his commitment. Additional backing came from firms including Lux Capital, Google Ventures, and Catalio. Flourish has since raised approximately $500 million and achieved a reported valuation of $2.5 billion. The company now faces the challenge of proving that a fundamentally different approach to AI can outperform existing architectures.
The founders argue that modern AI has become increasingly inefficient. Current frontier models require enormous data centers, thousands of advanced chips, and massive amounts of electricity to train and operate. By contrast, the human brain consumes roughly 20 watts of power while performing tasks such as language understanding, reasoning, learning, and sensory processing. Humans also learn from relatively small amounts of information. A child can acquire language after hearing only a fraction of the data that AI systems require. Flourish believes this difference suggests that biology contains important computational principles that current AI models have not yet discovered.
A major focus of the company’s research is the study of cortical columns. Cortical columns are small, vertically organized groups of neurons found throughout the cerebral cortex, the outer layer of the brain responsible for perception, reasoning, memory, and decision-making. Each cortical column contains thousands of interconnected neurons arranged in layers. Although scientists are still debating exactly how they function, many believe these columns act as fundamental information-processing units that repeat throughout the cortex. Different regions of the cortex process vision, hearing, touch, language, and other functions, yet they all appear to use a similar column-based structure. Because of this repeating design, some neuroscientists describe cortical columns as the “canonical computational unit” of the brain. Flourish researchers hope that understanding how these columns process information, learn, and communicate could reveal new algorithms that make AI systems far more efficient than today’s transformer-based models.
To investigate these ideas, Flourish has assembled a team of neuroscientists and AI researchers who work closely together. The company plans to conduct original laboratory experiments using advanced equipment such as electron microscopes to study neural circuits at multiple scales. Researchers will examine everything from molecules and synapses to entire networks of cells. Insights from these experiments will guide the development of new AI models, while computational models may also help scientists better understand biological systems.
The company is already exploring commercial applications based on recent neuroscience findings. One promising area involves a hippocampus-inspired memory system that could allow AI models to learn continuously rather than remaining fixed after training. Flourish is also developing algorithms designed to run on consumer devices and is reportedly discussing hardware implementations with major chip manufacturers. These technologies could provide near-term products while supporting the broader mission of creating AI systems that more closely resemble the capabilities of the human brain.
Although the project remains highly speculative, its potential impact is enormous. If Flourish succeeds in discovering the fundamental algorithms underlying biological intelligence, it could dramatically reduce AI energy consumption, enable continuous learning, and reshape the future of artificial intelligence. Investors and researchers acknowledge the risks, but they also recognize that success could transform both neuroscience and AI, potentially making current data-center-heavy approaches far less important.
https://www.wired.com/story/jeff-bezos-is-funding-a-wild-hunt-for-the-brains-core-algorithm