A Multi-Agent System for Automating Scientific Discovery

Summary

Robin is a multi-agent system from FutureHouse that integrates literature-search agents with data-analysis agents to run the full hypothesis → experiment-proposal → result-interpretation → revised-hypothesis loop. The system was deployed on a real biomedical problem — dry age-related macular degeneration (dAMD), the leading cause of blindness in the developed world — and autonomously proposed enhancing retinal pigment epithelium (RPE) phagocytosis as a therapeutic strategy. Robin then identified and validated ripasudil (a clinically-used ROCK inhibitor never previously proposed for dAMD) as a promising candidate, and proposed a follow-up RNA-seq experiment that revealed ABCA1 upregulation as a possible novel target.

Contribution

The first reported AI system to autonomously discover and wet-lab validate a novel therapeutic candidate within an iterative lab-in-the-loop framework. All hypotheses, experimental plans, data analyses, and figures in the main text were produced by Robin — humans executed the bench experiments, but the intellectual loop is end-to-end automated. This goes beyond prior AI-scientist systems (AI Scientist, EvoScientist, ARIS, data-to-paper) which automate the writing of a research paper but stop short of generating a validated biomedical discovery.

Method

System paper with a single, deep biomedical case study. Authors built a multi-agent harness combining literature-search and data-analysis roles; deployed it on the dAMD problem; performed the proposed wet-lab experiments; iterated. Evaluation is by the scientific outcome itself (a novel, mechanistically-validated drug candidate) rather than by benchmark metrics.

Relevance to RISE

Among the most important recent agentic-research-pipeline papers for the RISE catalog. Distinguishing features vs. other end-to-end AI scientists in the catalog:

• Lab-in-the-loop validation. Where yang2026aris, evoscientist2026techreport, and ifargan2024datatopaper validate against benchmarks or recapitulate known findings, Robin validates against reality (a wet-lab assay).
• Discovery as the evaluation criterion. The system is evaluated by whether it produced a novel, real, mechanistically-grounded therapeutic candidate — a much harder bar than "novelty score from an LLM judge".
• Sociotechnical implications. A concrete, peer-reviewed (Nature) demonstration that the agentic-discovery thread can produce outcomes that matter outside the AI literature itself. Useful for the introduction of the RISE position paper.

Critique / open questions

A single case study — the result is striking but generalisability to other biomedical (or non-biomedical) discovery problems is open. Lab-in-the-loop is semi-autonomous: humans still run the wet experiments, set up the assays, and presumably curate the candidate list at decision points. The paper does not report how many ripasudil-class false positives Robin generated before this one was validated — a base-rate analysis would be needed to compare against domain experts on the same task. The "all hypotheses and figures produced by Robin" claim is strong; readers should look at the methods carefully to understand exactly which decision points involved human input.

Key quotes

"Here, we introduce Robin, the first multi-agent system capable of fully automating the key intellectual steps of the scientific process. By integrating literature search agents with data analysis agents, Robin can generate hypotheses, propose experiments, interpret experimental results, and generate updated hypotheses, achieving a semi-autonomous approach to scientific discovery."

"By applying this system, we were able to identify a novel treatment for dry age-related macular degeneration (dAMD), the major cause of blindness in the developed world. Robin proposed enhancing retinal pigment epithelium phagocytosis as a therapeutic strategy, and identified and validated a promising therapeutic candidate, ripasudil."

"As the first AI system to autonomously discover and validate a novel therapeutic candidate within an iterative lab-in-the-loop framework, Robin establishes a new paradigm for AI-driven scientific discovery."