When will a FutureHouse Robin-style multi-agent system produce a candidate that reaches clinical trials?
Chance it happens
67%Belief the predicted event happens at all — every “yes / by-a-date” outcome added together. “Other” is tracked as its own third slice.
Outcomes
Resolution criteria
A candidate whose origin is attributed to a lab-in-the-loop multi-agent AI system enters a registered Phase 1. Resolution horizon: 2031-12-31. Source: publication, registry.
Description
FutureHouse is a nonprofit AI research lab (backed by Eric Schmidt) building AI "scientists" — autonomous agents that read the literature, form hypotheses, and help design experiments. Robin is their multi-agent system for end-to-end scientific discovery: it chains specialized agents together — Crow and Falcon, which search and read the research literature, and Finch, which analyzes experimental data — in a loop with real bench experiments. The AI proposes an idea, a human runs the wet-lab test, and the results feed back in to sharpen the next idea. This "lab-in-the-loop" pattern is what the market means by a Robin-style system. In its first public demonstration (announced May 2025, published in Nature in May 2026) Robin proposed ripasudil — a Rho-kinase (ROCK) inhibitor already sold as a glaucoma eye-drop — as a new treatment for dry age-related macular degeneration (dAMD, a leading cause of age-related blindness), running the whole hypothesis-to-candidate cycle in about 2.5 months. Background: https://www.futurehouse.org/research-announcements/demonstrating-end-to-end-scientific-discovery-with-robin-a-multi-agent-system So the whole thing in plain words: when will a drug candidate whose discovery is genuinely credited to this kind of multi-agent AI system — an AI running the reasoning while humans run the experiments — actually reach a registered human clinical trial (Phase 1, the first-in-human safety study)? The bar is deliberately high: the AI system has to be credited as the origin of the candidate, not merely used as one tool along the way. Resolution horizon: 2031.
Evidence & context
Trials, publications and reports our research found on this prognosis. Each links to the original source.
This is a 2026 Nature publication describing Robin, a lab-in-the-loop multi-agent AI system for scientific discovery. It bears directly on the forecast because it reports AI-originated therapeutic candidate discovery and in vitro validation, showing substantial progress toward the resolution criterion even though it does not itself report entry into a registered Phase 1 trial.
This is a FutureHouse research announcement describing Robin as a multi-agent system for biological discovery and stating the work was published in Nature on May 19, 2026. It is not evidence of a candidate entering Phase 1, but it is directly relevant capability evidence for forecasting whether Robin-style systems could eventually generate a clinical-trial candidate.
This page discusses a 2025 paper from Oxford and FutureHouse describing Robin, a multi-agent system aimed at automating scientific discovery in medicine, specifically drug repurposing. It is relevant because it provides direct evidence of the existence and scope of a FutureHouse Robin-style lab-in-the-loop AI system, which informs how close such systems may be to producing candidates that could enter Phase 1.
This is a Chemical & Engineering News article describing newly introduced agent-based scientific discovery systems, including FutureHouse's tools, and stating they can generate hypotheses, design experiments, and analyze data. It is relevant as background evidence on the maturity and capabilities of Robin-style multi-agent AI systems, though it does not report a candidate entering Phase 1.
This is FutureHouse's official homepage describing the organization as a nonprofit building AI agents to automate research in biology and other complex sciences. It is relevant as direct evidence that the organization is actively pursuing multi-agent AI-for-science work, though it does not report a drug candidate or clinical trial milestone.
This appears to be a LinkedIn announcement by FutureHouse founder Samuel G. Rodriques claiming an AI system discovered a new treatment for dry AMD. It is relevant because it is direct evidence that a Robin-style multi-agent drug-discovery effort is producing therapeutic candidates, even though the page does not show entry into a registered Phase 1 trial.
This appears to be a LinkedIn post discussing FutureHouse's multi-agent AI system Robin and claiming it found a treatment candidate for dry age-related macular degeneration (dAMD). It is relevant because it is direct evidence that a Robin-style multi-agent system is being presented as producing therapeutic candidates, which bears on the likelihood of such a candidate eventually reaching Phase 1.