May 19, 2026 ScienceUniting biological toolkits for a new approach to ALS ShareRitu Raman at MIT and Ryan Flynn at Boston Children's Hospital approach human biology with very different toolkits, but Co-Scientist is bridging their labs. Raman, a mechanical engineer, builds living nerve and muscle tissues to model diseases that affect voluntary movement. Her husband Flynn, a chemical biologist, maps RNA on the surface of cells to see how it influences cellular communication and how pathogens invade.When Raman decided to investigate ALS, which was outside of her usual domain, she faced a sprawling, contradictory literature that would usually take months to grasp. Co-Scientist compressed that work, quickly helping Raman interrogate the evidence in relation to her tissue model, turn ideas into testable hypotheses, and rank potential directions in accordance with the trade-offs labs actually face, such as feasibility and potential risk–reward.But Co-Scientist’s best leads came with a catch: they involved what happens at the surface of cells, where much of their communication is mediated. Raman could manipulate tissues and measure outcomes, but decoding the molecular interactions driving those signals was outside her area of expertise.That gap became the catalyst for collaboration. Raman brought her new research directions to Flynn, and the pair used Co-Scientist iteratively, combining its best ideas into creative research pathways that united their distinct toolkits. To develop new therapies, their hunt is now on for novel RNA-based mechanisms—and potentially RNA-based drugs—that could be used to target ALS.Slide 1 of 2Science is a team sport. Co-Scientist can’t do science by itself, and I can’t do it all by myself either. It helps me structure my thoughts, so I know what to ask of other experts and collaborators. Associate Professor Ritu Raman Massachusetts Institute of Technology You take pieces of Co-Scientist’s top-ranked concepts, and synthesize them into a new thing. It helps you take good, logical ideas, shift them a few degrees, and turn them into something even more creative. Associate Professor Ryan Flynn Boston Children’s Hospital Related postsCo-Scientist: A multi-agent AI partner to accelerate researchMay 2026Science Learn more Uncovering repurposed medicines to fight liver fibrosisMay 2026Science Learn more Accelerating discovery of liver disease mechanismsMay 2026Science Learn more Opening new paths in aging researchMay 2026Science Learn more Finding the molecular switches behind new infectious diseasesMay 2026Science Learn more Fast-tracking genetic leads to reverse cellular agingMay 2026Science Learn more
Co-Scientist
Driving creative collaboration in research
Related
ResearchLearning When to Trust via Selective Context Preference Optimization
Language models increasingly condition their answers on external signals, and a single misleading one can turn a correct answer wrong. The obvious remedy, training models to resist such signals, hides a failure mode: a model that ignores all context looks robust yet is useless when the context is worth trusting. We recast the problem as selective trust and introduce MIST, a human-annotated benchmark that renders each reasoning item under four matched conditions (clean, misleading, correct-context, and irrelevant-context), together with SC2W, a paired metric counting how often a misle
Learning When to Trust via Selective Context Preference Optimization
Language models increasingly condition their answers on external signals, and a single misleading one can turn a correct answer wrong. The obvious remedy, tr...
ResearchTracing the Heart: An Evidence-Linked Pipeline for Heart-Failure Feature Engineering
Electronic health record (EHR) feature engineering is a major bottleneck in clinical research and AI, accounting for 39-45% of data scientists' workload. This is especially pronounced in heart failure, which affects an estimated 6.7 million U.S. adults and requires integrating fragmented EHR data with disease-specific, guideline-based clinical reasoning. Existing rule-based and large language model (LLM)-based approaches offer only partial automation with limited maintainability and evidence traceability. We developed the Nimblemind Multi-Agent System (nMAS), an evidence-linked, rubr