Put The Lab In The Loop

2026-07-17 · Source: The Batch | DeepLearning.AI | AI News & Insights - www.deeplearning.ai · Field: Science & Research — Artificial Intelligence & Machine Learning, Health & Medical Research, Robotics & Autonomous Systems · Depth: Advanced, short

Summary

Ali Essam Ghareeb, Benjamin Chang, and colleagues from FutureHouse, University of Oxford, and Fordham University released Robin, an open-source AI agent designed to propose new medical uses for established drugs. Robin identified two drugs, Y-27632 and Ripasudil, that address a biological mechanism behind dry age-related macular degeneration (dAMD), a leading cause of impaired vision. The agent operates by iteratively identifying disease mechanisms, designing experiments, finding existing commercially available drugs, and analyzing lab results. It utilizes OpenAI's GPT-4-mini for language processing and Claude 3.7 Sonnet for ranking experimental designs and drug reports. While Robin itself is available under an Apache 2.0 license, it relies on proprietary literature-search agents Crow and Falcon (research-use only) and the open-source data-analysis agent Finch. Experiments on isolated human eye cells showed Y-27632 increased RPE phagocytosis by nearly 2x, and Ripasudil, an approved glaucoma drug in Japan, by 1.89x (human analysis 1.75x).

Key takeaway

For research scientists and AI/ML directors focused on drug discovery, Robin demonstrates a powerful model for accelerating research. You should consider integrating iterative AI agent systems into your R&D pipelines to autonomously generate hypotheses, design experiments, and analyze results. This approach can significantly reduce the time and cost associated with identifying new therapeutic uses for existing drugs, streamlining your path to validated candidates.

Key insights

AI agents can autonomously propose drug repurposing candidates and iterate on experimental results with minimal human input.

Principles

• Iterative AI-human loops accelerate scientific discovery.
• Agentic systems can integrate diverse AI models for complex tasks.
• Repurposing existing drugs offers efficiency in development.

Method

Robin identifies disease mechanisms, designs experiments, finds existing drugs, and analyzes human-run lab results in an iterative loop until a candidate is satisfactory.

In practice

• Use Robin for drug repurposing hypothesis generation.
• Integrate AI agents for literature review and data analysis.
• Apply agentic workflows to other scientific research areas.

Topics

• AI Agents
• Drug Repurposing
• Medical Research Automation
• Dry AMD Treatment
• Scientific Discovery
• Human-in-the-Loop AI

Code references

• Future-House/robin
• Future-House/finch

Best for: AI Scientist, Research Scientist, Director of AI/ML

Related on AIssential

• Two AI-based science assistants succeed with drug-retargeting tasks — AI-powered agentic systems can accelerate scientific discovery by synthesizing vast literature and generating testable hypotheses.
• Teams of AI agents boost speed of research — AI agent teams can rapidly generate plausible scientific hypotheses and identify drug candidates, significantly accelerating early-stage research.
• He Raised $70M to Cure Every Disease With AI — AI agents, particularly specialized models, significantly accelerate scientific discovery by removing human talent bottlenecks and enabling high-throughput reasoning.
• Using AI to Develop New Drugs — AI is fundamentally transforming drug discovery and development processes.
• Stanford researchers will discuss their agentic 'scientists' that are on course to reshape drug discovery at VB Transform 2026 — Agentic AI systems with hierarchical orchestration and continuous context can significantly improve the efficiency of complex, multi-stage processes like drug discovery.
• How AI helps scientists design the next generation of medicines — AI fundamentally accelerates and expands complex biologic medicine design, enabling previously impossible discoveries.

See Counsel's argued verdicts on the open AI decisions leaders are weighing →

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by The Batch | DeepLearning.AI | AI News & Insights - www.deeplearning.ai.