What FutureHouse does

FutureHouse is a 501(c)(3) nonprofit AI-for-science research lab founded in 2023 and headquartered in San Francisco, with a stated mission of building AI Scientists to automate and accelerate fundamental research in biology and adjacent complex sciences. The organization operates the AI-for-Science Independent Postdoctoral Fellowship as its flagship program, pairing early-career researchers with advanced AI tools and academic co-advisors at leading institutions including the Broad Institute, Stanford, Caltech, MIT, and Harvard. FutureHouse also publishes open research, releases open-source software, and offers a free self-paced tutorial series to help academic scientists integrate AI into their research workflows.

The technical portfolio centers on Robin, a multi-agent orchestration system that integrates specialized agents (Crow for literature synthesis, Falcon for chemical analysis, Finch for data analysis) to autonomously generate hypotheses, design experiments, analyze data, and validate findings. Supporting systems include PaperQA2, a retrieval-augmented generation agent that achieves expert-level performance on scientific literature tasks; DISCO, a multimodal diffusion model for simultaneous protein sequence and 3D structure design; OXtal, a generative model for organic crystal structure prediction; WikiCrow for automated scientific article synthesis; and the Aviary training framework for scientific language agents. FutureHouse has also released benchmarks (LAB-Bench, BixBench) that establish community evaluation standards. In May 2026, Robin's end-to-end autonomous discovery of ripasudil as a therapeutic candidate for dry AMD was published in Nature, marking the first validated AI-generated therapeutic hypothesis for a human disease.

FutureHouse generates no commercial product revenue and is funded entirely through philanthropic grants and donations. Initial two-year funding came from Eric and Wendy Schmidt, with additional support from OpenPhilanthropy, the National Science Foundation, and the AI Safety Institute. The organization accepts donations through its website and donor-advised funds and offers a Fellowship Sponsorship program for individuals and organizations wishing to support specific postdocs. A commercial counterpart, Edison Scientific, was spun out in 2023 to bring FutureHouse's AI research outputs to market under a separate governance structure, with the two entities sharing office space but operating independent missions.

Name

FutureHouse

Legal name

Future House, Inc.

Website

https://futurehouse.org

Company type

Private

Founded year

2023

Operating status

Operating

Headcount range

1–10 employees

Short description

FutureHouse is a 501(c)(3) nonprofit AI research lab developing multi-agent systems, generative models, and benchmarks to automate scientific discovery in biology, serving academic researchers through postdoctoral fellowships and open-source tool releases.

Ownership category

akta.pro rank

Product category

Scientific Discovery AI Software

NAICS

Research and Development in the Physical, Engineering, and Life Sciences (54171), Scientific Research and Development Services (5417), Research and Development in Biotechnology (except Nanobiotechnology) (541714)

SIC

Services-Commercial Physical & Biological Research (8731), Services-Prepackaged Software (7372)

akta.pro primary industry

AI/ML Platforms for Drug Discovery & Experiment Optimization (HLAGAJAO)

akta.pro secondary industries

Scientific Data Management & Data Platforms (SDMS/Data Lakehouse for R&D) (HLAGAJAC), Scenario Planning, Forecasting Platforms & Planning Intelligence (HDAAAHAL)

Keywords

• AI research agents
• Scientific discovery automation
• Multi-agent systems
• Computational biology research
• AI for science

Headquarters

HQ city

San Francisco

HQ country

United States

HQ region

North America

Offices1 record

Markets served

GTM type

B2B

Offering type

Software

Cost components

Personnel, Technology or R&D, Operations, Marketing or Sales, Infrastructure

Revenue model

1. Philanthropic Funding: FutureHouse is funded through grants and donations from foundations and philanthropists. Initial funding came from Eric and Wendy Schmidt for the first two years, with additional support from OpenPhilanthropy, National Science Foundation, and AI Safety Institute. The organization accepts donations through its website and donor advised funds.
2. Fellowship Sponsorship: Organizations and individuals can sponsor FutureHouse Fellows, providing resources for early-career scientists to conduct AI-accelerated research. Sponsors receive association with cutting-edge research but the nonprofit does not sell products or services.

Go-to-market motion1 record

Distribution channels4 records

Marketing channels7 records

Core offering

FutureHouse is a 501(c)(3) nonprofit AI-for-science research lab that develops AI agents and multi-agent systems to automate scientific discovery in biology and complex sciences. Its portfolio includes Robin (a multi-agent orchestration system for end-to-end discovery), PaperQA/PaperQA2 (scientific literature RAG agent), DISCO (enzyme design model), OXtal (crystal structure prediction), and open benchmarks (LAB-Bench, BixBench, Aviary), all distributed openly to the scientific community alongside a postdoctoral fellowship program.

Product overview

FutureHouse is a nonprofit AI-for-science research lab developing a portfolio of AI systems to automate and accelerate scientific discovery in biology. The core product is Robin, a multi-agent orchestration system that integrates specialized agents (Crow, Falcon, Finch) for end-to-end discovery from hypothesis to validation. Supporting products include PaperQA2 for literature synthesis, DISCO for protein design, and OXtal for crystal structure prediction. The portfolio also includes benchmarks (LAB-Bench, BixBench) and training frameworks (Aviary) for evaluating and developing scientific AI capabilities. The AI Scientist concept represents the overarching vision: a system that takes broad scientific questions and autonomously generates and validates novel discoveries.

Differentiator

Problem solved

Functional benefit

Products and services

• Robin Multi-Agent System Multi-agent AI system that autonomously generates hypotheses, designs experiments, analyzes data, and validates findings for end-to-end scientific discovery; built for academic and industry biology researchers.
• PaperQA / PaperQA2 Retrieval-Augmented Generation agent for scientific literature that achieves expert-level performance on PubMedQA and other benchmarks by combining literature search, evidence gathering, and answer synthesis; available open-source on GitHub for researchers.
• FutureHouse AI Agent Platform (Crow, Owl, Falcon, Phoenix, Finch) Suite of specialized AI agents hosted on a web platform: Crow for literature synthesis, Owl for scientific literature search, Falcon for molecule evaluation, Phoenix for chemical synthesis design, and Finch for complex data analysis; built for biologists and chemistry researchers.
• DISCO (Diffusion for Sequence-structure CO-design) Multimodal generative model that simultaneously designs protein amino acid sequence and 3D structure co-folded with arbitrary biomolecules; used by enzyme designers and synthetic biology researchers for new-to-nature chemistry.
• OXtal 100M-parameter all-atom diffusion model that predicts 3D organic crystal packings from 2D molecular graphs in seconds on a single GPU, replacing brute-force physics simulations; used by chemists and materials scientists.
• WikiCrow Automated system that synthesizes cited Wikipedia-style summaries from scientific literature for protein-coding genes; available open-source for the scientific community.
• LAB-Bench (Language Agent Biology Benchmark) Benchmark of 2,457 questions across 8 categories that measures language model capabilities for biology research tasks such as literature extraction, database retrieval, sequence manipulation, and protocol design; available open-source.
• BixBench Benchmark of 53 real-world bioinformatics analytical scenarios with 296 research questions that evaluates AI agents on autonomous data analysis and insight generation over biological datasets; available open-source.
• Aviary Environment framework for training language agents on challenging scientific biology tasks, providing tools for protein engineering and literature summarization; available open-source.
• AI-for-Science Postdoctoral Fellowship One-year postdoctoral fellowship (with possible extension) that pairs early-career scientists with FutureHouse AI tools, compute, engineering support, and academic co-advisors at leading institutions; intended for PhD graduates and postdocs pursuing AI-accelerated biology research.
• AI Tutorial Series Free self-paced online tutorial series covering LLMs, agents, and practical AI implementation; targeted at scientists from non-computational backgrounds who want to integrate AI into research workflows.

Quantifiable outcome

• A single day of AI Scientist can accomplish what human scientists accomplish in 6 months
• +4 more outcomes

Segments3 records

Ideal customer profiles3 records

Technology focussed Yes

API detail

Has API

No

API docs

API detail

Core technology

AI maturity

App detail

Integration2 records

AI capability9 records

Feature10 records

Partnerships

Seven partnerships are on record, tiered core and minor.

• Edison ScientificcoreStrategic or Co-development Partner · 1 January 2023Edison Scientific is the commercial spinout of FutureHouse, bringing the AI technology developed at FutureHouse to market. FutureHouse is a nonprofit focused on fundamental research for public good while Edison Scientific commercializes AI tools. The two organizations share office space but have separate missions, operations, and governance. FutureHouse has access to Edison's technology.
• Broad InstitutecoreStrategic or Co-development PartnerFei Chen at Broad Institute serves as co-advisor for FutureHouse Fellow Blake Lash, who is developing AI tools to decode how parasitic worms suppress the human immune system.
• Stanford UniversitycoreStrategic or Co-development PartnerJesse Engreitz at Stanford University co-advises FutureHouse Fellow Philine Guckelberger, who is using AI to decode how 3D DNA structure controls gene activity in cancer.
• California Institute of Technology (Caltech)coreStrategic or Co-development PartnerFrances Arnold at Caltech serves as co-advisor for FutureHouse Fellow Chenghao Liu, who is building AI systems to design new enzymes. DISCO research collaboration involved Caltech researchers including Frances Arnold (co-author and Nobel laureate).
• Massachusetts Institute of Technology (MIT)coreStrategic or Co-development PartnerSergey Ovchinnikov at MIT serves as co-advisor for FutureHouse Fellow Sarah Gurev, who specializes in protein design and virology.
• Harvard T.H. Chan School of Public HealthcoreStrategic or Co-development PartnerPardis Sabeti at Harvard co-advises FutureHouse Fellow Laura Luebbert, who is developing machine learning methods for infectious disease discovery.
• ScienceMachineminorStrategic or Co-development PartnerScienceMachine partnered with FutureHouse to develop BixBench, a benchmark for evaluating AI agents on bioinformatics tasks.

Scale indicators8 records

Recent moves6 records

Expansion highlights6 records

Emerging players

• EvolutionaryScale: AI research lab developing frontier foundation models for biology, including protein language models. Emerging player with comparable focus on AI-native biological discovery.
• Cradle Bio: AI protein design company using generative models to engineer novel proteins for biotech applications. Direct overlap with FutureHouse's DISCO multimodal generative model for protein sequence and structure co-design.

Direct peers

• Insilico Medicine: AI-driven drug discovery company that uses generative AI and reinforcement learning to identify novel targets and design molecules. Competes with FutureHouse in end-to-end AI-driven therapeutic discovery.
• Recursion Pharmaceuticals: TechBio company combining high-throughput biology with AI/ML to discover medicines at scale. Competes with FutureHouse in applying AI to industrial-scale drug discovery and biology research.
• Isomorphic Labs: Alphabet/Google DeepMind subsidiary applying AI to drug discovery and biomedical research. Builds AlphaFold-derived platforms for protein structure prediction and therapeutic design, directly overlapping with FutureHouse's Robin, DISCO, and OXtal capabilities.
• Atomwise: AI-driven small molecule drug design company using structure-based deep learning models. Comparable to FutureHouse's DISCO and OXtal for protein/crystal design and small molecule discovery.
• Edison Scientific: Commercial spinout of FutureHouse that commercializes the AI tools developed at the nonprofit, including scientific discovery agents. Shares office space and has access to FutureHouse's research, but operates independently with a for-profit mission.

Broad incumbents

• Google DeepMind: Frontier AI research lab with significant investment in scientific applications including AlphaFold, AlphaMissense, and other biology-focused AI systems. Broad incumbent that could subsume many of FutureHouse's capabilities.
• Anthropic: Frontier AI lab with general-purpose models increasingly applied to scientific reasoning and literature analysis. Broad incumbent with potential to compete in scientific agent capabilities.

Others

• Allen Institute for AI (AI2): Nonprofit AI research institute (Semantic Scholar partner of FutureHouse) that develops open AI models and scientific literature tools. Thematically aligned nonprofit AI-for-science peer and ecosystem partner.

Market position

Competitive moat5 records

Key risks6 records

Key highlights6 records

Customer concentration

• Xx.com/futurehousesf
• LinkedInlinkedin.com/company/futurehouse

Revenue estimate

Valuation estimate

Number of profiles

Profiles4 records

Funding overview

Funding rounds

Investors

Funding detail is available on the Subscription and Enterprise plan.Contact sales →

M&A

Investments

M&A and investment is available on the Subscription and Enterprise plan.Contact sales →

Frequently asked questions about FutureHouse