We're building an AI scientist.
FutureHouse is a nonprofit research lab using AI to accelerate the science that improves human health.
The Challenge
Solving biology requires scale.
There are 20,000 genes in the human genome. Tens of thousands of cell types in the brain. Thousands of diseases — and not enough scientists to research all of them thoroughly.
Biology's biggest problems aren't limited by ambition. They're limited by the number of trained minds that can read the literature, form hypotheses, design experiments, and interpret the results. AI changes that equation.
What we do
Pairing exceptional scientists with advanced AI agents.
FutureHouse is a nonprofit independent research organization based in San Francisco, founded in 2023 by Sam Rodriques and Andrew White. We build and validate AI systems for scientific discovery — primarily in biology right now — and run initiatives that put those systems in the hands of exceptional scientists.
Our flagship program today is the AI-for-Science Independent Postdoctoral Fellowship, which pairs early-career researchers with advanced AI tools and academic co-advisors at leading research institutions. We also develop training resources to help scientists in academia integrate AI into their own research workflows.
FutureHouse focuses on use-inspired basic research — fundamental science motivated by real-world problems. Edison Scientific, a for-profit company that spun out of FutureHouse in 2025, commercializes the AI tools. FutureHouse has access to Edison's technology and remains focused on its nonprofit mission: accelerating discovery for the public good.
Our impact
The promise of an AI Scientist is closer than even we imagined.
In our first two years, our team produced results that demonstrate what’s possible when AI meets rigorous science.
PaperQA2, the first AI agent to achieve superhuman performance on scientific literature tasks, outperforms PhD-level researchers in head-to-head evaluations. Aviary made it possible to train language agents on real scientific tasks for the first time, from literature research to DNA reasoning to protein design. Both are open-source and available for the broader research community to build on.
Robin, the multi-agent system developed at FutureHouse, autonomously generated hypotheses, designed experiments, analyzed data, and identified ripasudil — an existing glaucoma drug — as a promising therapy for dry age-related macular degeneration, a leading cause of blindness. The entire process from concept to paper took 2.5 months.
ether0 is a reasoning model trained specifically for chemistry with reinforcement learning. It outperforms leading general-purpose AI models on molecular design tasks that can lead to pharmaceutical discoveries. It is open-source and available for any researcher to use.
FutureHouse has built evidence-based AI benchmarks, shaping the field’s standards for measuring AI performance on scientific tasks. For example, we developed LAB-Bench, a benchmark of practical research tasks, BixBench for open-ended bioinformatics tasks, and we published an independent audit that identified errors in a widely used AI evaluation.
In 2025, we launched Kosmos, our first AI Scientist capable of conducting complex analyses autonomously, as confirmed by expert validation. Kosmos conducts six months of research in a single day, and we estimate 80% of its findings to be accurate based on our own evaluations. Edison Scientific is bringing this technology to the world, and FutureHouse is using it to make basic science breakthroughs.
FutureHouse develops the next generation of AI-fluent scientists by hosting outstanding early career scientists through our Post-doctoral Fellowship program. Fellows lead their own research projects at the intersection of AI and biology, co-advised by leading academic scientists.
Our space
Surrounded by and engineered for innovation
FutureHouse operates from a converted warehouse in San Francisco's Dogpatch neighborhood, surrounded by companies including IndieBio and Y Combinator. The space includes a 3,000 sq. ft. wet lab, an event space designed by Frozen Feathers studio, and an outdoor area for gathering and collaboration.
It's intentional. FutureHouse isn't a traditional research institute, and the space reflects: open, energetic, built for a team that moves between computation and the lab bench.
Engage with us
Partner with FutureHouse
FutureHouse supports focused giving through its fellowship program and research initiatives. If your organization is interested in accelerating discovery in a specific disease area or scientific domain, we'd welcome the conversation.
We're a 501(c)(3) nonprofit AI-for-science research lab in San Francisco. Our mission is to build an AI Scientist to automate and accelerate fundamental science that improves human health.
FutureHouse is funded through philanthropic partnerships and grants. Our first two years were made possible by Eric and Wendy Schmidt, and we have received additional support from OpenPhilanthropy, the National Science Foundation, and the AI Safety Institute, among others.
FutureHouse and Edison Scientific are two independent organizations. FutureHouse is a nonprofit focused on fundamental research for the public good. Edison Scientific is a for-profit company that commercializes AI tools for science. FutureHouse has access to Edison's technology. The two organizations share office space but have separate missions, operations, and governance.
The AI systems originally developed under the FutureHouse name are now part of Edison Scientific's platform.
