Scientific Software Development

Excited to share our latest work published in Nature Catalysis: 𝗔𝘂𝘁𝗼𝗻𝗼𝗺𝗼𝘂𝘀 𝗰𝗮𝘁𝗮𝗹𝘆𝘀𝗶𝘀 𝗿𝗲𝘀𝗲𝗮𝗿𝗰𝗵 𝘄𝗶𝘁𝗵 𝗵𝘂𝗺𝗮𝗻–𝗔𝗜–𝗿𝗼𝗯𝗼𝘁 𝗰𝗼𝗹𝗹𝗮𝗯𝗼𝗿𝗮𝘁𝗶𝗼𝗻 Read the article here: https://lnkd.in/dCfrNf5S Catalysis drives modern chemical manufacturing and sustainability, but traditional discovery remains slow and resource-intensive. In this article, we discuss how self-driving laboratories that couple automation, AI, and high-throughput experimentation can accelerate catalysis research, while ensuring rigorous human oversight. We present a forward-looking roadmap for human–AI–robot collaboration in catalysis, highlighting opportunities to integrate data-centric infrastructure, scalable automation, and trustworthy AI for accelerated and reproducible discovery. Congratulations to Negin Orouji and all co-authors, Jeffrey Bennett, Richard Canty, Long Qi, Shijing Sun, Paulami Majumdar, Chong Liu, Nuria Lopez, Neil Schweitzer, John Kitchin, and Hongliang Xin. Thanks to the National Science Foundation (NSF) for supporting this work. #AutonomousScienceAndEngineering #Selfdrivinglabs

AI can design 10,000 new molecules before lunch. Your synthesis robot? Maybe 20. Here’s what’s changing. Most automated chemistry platforms today operate in one of two modes: high-throughput experimentation (optimizing one reaction hundreds of ways) or library synthesis (building similar compounds around a shared scaffold). Both are designed to repeat, not to explore. But AI-driven molecular design has changed the game. Models now generate diverse candidates that span disconnected regions of chemical space—and that diversity breaks traditional automation. A recent preprint from Franck Le Vaillant et al. at iktos proposes a new paradigm: cluster synthesis. Instead of batching identical reactions, their robotic platform groups different chemistries that share compatible reaction conditions such as temperature and time ranges. Here’s how their system turns this idea into reality: 🔹Design with constraints. An AI generates molecules only from reactions the robot can execute. 🔹Route planning: A retrosynthesis planner proposes synthetic routes, based on reaction templates to maximize compatibility. 🔹Tactical scheduling: An optimization algorithm finds the best clustering based on overlapping conditions and available ingredients and reactor capacity. This approach enables parallel running of different reactions such as Suzuki couplings, reductive aminations, SNAr reactions, and peptide couplings. In just three robotic campaigns, the team synthesized 135 molecules spanning 27 reaction types. The results are impressive: a 72 % success rate and a synthesis process 2-4 times faster than conventional setups. More broadly, it reframes what automation can be, turning a synthesis robot from a single-purpose tool to a shared infrastructure for discovery. 📄 Thinking Outside the Library: Cluster Synthesis of Diverse Molecules on a Single Robotic Platform, ChemRxiv, October 8, 2025 🔗 https://lnkd.in/edG8Rzvu

Low-Cost RoboChem System Democratizes AI-Driven Chemical Research A new breakthrough in laboratory automation is set to expand access to advanced chemical research tools. The RoboChem Flex system, developed by researchers at the University of Amsterdam, delivers an AI-powered synthesis optimization platform at an estimated cost of just five thousand dollars, dramatically lowering the barrier to entry for autonomous experimentation. Unlike traditional automated chemistry systems that are often limited to well-funded institutions, RoboChem Flex is designed with modular components and open accessibility. Laboratories can build and deploy the system using published specifications, enabling widespread adoption. The platform integrates artificial intelligence to optimize chemical synthesis processes, while still allowing for human oversight and intervention where needed. This hybrid model ensures both efficiency and scientific control. The system’s versatility allows it to adapt to a wide range of experimental setups, making it suitable for both small research groups and larger institutions. By automating repetitive and data-intensive tasks, it accelerates discovery cycles and improves reproducibility. Researchers can explore more variables in less time, enhancing the overall pace of innovation in chemical sciences. The broader implications are significant for global research equity and scientific progress. By reducing cost and complexity, RoboChem Flex enables a more distributed model of innovation, where smaller labs can contribute meaningfully to advanced research. This democratization of tools could accelerate breakthroughs across pharmaceuticals, materials science, and energy, reinforcing the role of accessible technology in driving the next wave of scientific discovery. I share daily insights with tens of thousands followers across defense, tech, and policy. If this topic resonates, I invite you to connect and continue the conversation. Keith King https://lnkd.in/gHPvUttw

An AI system just completed 6 months of research work in a single day. Allegedly. Scientists have long dreamed of automating the tedious cycles of literature review, hypothesis generation, and data analysis that define modern research. While recent AI agents have shown promise in specific tasks, they've all hit the same wall: they lose coherence after just a handful of actions, limiting the depth and quality of their discoveries. Kosmos breaks through this limitation with a structured "world model" that acts as a shared memory system between specialized agents. One agent handles data analysis while another scours scientific literature - and both continuously update a central knowledge base that keeps them aligned over hundreds of parallel tasks. This architecture enables Kosmos to execute an average of 42,000 lines of code and read 1,500 full-length papers in a single 12-hour run, analyzing everything from metabolomics to materials science to statistical genetics. Independent experts confirmed 79.4% of Kosmos' statements as accurate, and collaborating research groups estimated that a single Kosmos run performed the equivalent of 6 months of their own research time. According to the authors, Kosmos doesn't just replicate known findings - it has made genuinely novel discoveries, including identifying a previously unknown mechanism of neuronal vulnerability in Alzheimer's disease. The system's structured approach ensures every claim traces back to either executable code or primary literature citations, making the entire discovery process transparent and reproducible. As datasets grow larger and more complex across scientific disciplines, tools like Kosmos demonstrate how AI can augment rather than replace human scientists - handling the computational heavy lifting while researchers focus on interpretation, experimental design, and directing investigations toward meaningful questions. ↓ 𝐖𝐚𝐧𝐭 𝐭𝐨 𝐤𝐞𝐞𝐩 𝐮𝐩? Join my newsletter with 50k+ readers and be the first to learn about the latest AI research: llmwatch.com 💡

🤔Do I think self-driving labs will be a part of the future? Not really.... Thanks to great discussions with my colleagues at North Carolina State University (and opportunity to think during the drive from Charlotte to Knoxville, courtesy of American Airways:)). When I talk to colleagues and observe how thinking is evolving across the scientific community, I don’t see autonomous labs as just a part of the future—I see them as everywhere. This isn’t a distant vision; it’s already happening. But there’s a catch. You can’t just buy a robot to pipette, synthesize, or measure and expect your science to transform. Automation only works when the whole process is designed to be cohesive. Imagine you need a gram of material for characterization—but your new robotic synthesis platform only produces milligrams or a compositional spread. That mismatch doesn’t help. In fact, it might even slow you down. However, that’s also where the magic starts. Once you align synthesis and characterization, and couple them intelligently, a new space opens up—where proxy data becomes powerful. Measurements that were once “not good enough” on their own become extremely informative when captured across large parameter spaces. Weak signals gain strength in context. In this sense, the future of self-driving labs is not about building perfect automation. It’s about building coordinated and context-aware workflows starting from synthesis-characterization pairs. This is how Mahshid Ahmadi lab started in 2019 - with Opentrons Labworks Inc. robot and Cytation optical reader. Automated film making and SPM followed and expanded these capabilities, but the seed was synthesis-spectroscopy pair. This transition can happen everywhere now. Total automation will still be needed in hazardous or extreme environments. But for most of us, the real opportunity is more exciting—and more democratic. Autonomous science will be everywhere. But it won’t be a robot takeover—it will be a new way of thinking. A lab where every experiment builds on the last, where humans and machines co-learn, and where discovery accelerates not because someone replaced the scientist—but because they freed them. Are we ready for this shift? Photo with Nina Balke , Raymond Unocic , and Kinga Unocic at NCSU.

Automation is transforming scientific research by tackling the bottlenecks in lab work. Germany's Jülich Research Centre has accelerated microorganism growth studies 100-fold by implementing robotic systems with high-throughput agar plates. Similarly, scientists worldwide are increasingly using automation to handle repetitive, high-precision tasks, from pipetting to sample analysis. In this video, TUM's Robotic Lab Assistant performs an autonomous, human-like pipetting task in a standard laboratory setup. It first analyzes the current setup, selects the appropriate pipette type for optimal pipetting performance, grasps it, and handles liquids according to the ISO 8655 standard procedure for a piston-operated volumetric apparatus. Not only does automation speed up results, but it also enhances accuracy and reproducibility—key factors in modern scientific advances. The future holds exciting potential for automation in academia, combining human expertise with robotic efficiency to fuel the next generation of discoveries. Read more: https://lnkd.in/dshXVtEw

"The second area is 'robot scientists', also known as 'self-driving labs'. These are robotic systems that use AI to form new hypotheses, based on analysis of existing data and literature, and then test those hypotheses by performing hundreds or thousands of experiments, in fields including systems biology and materials science. Unlike human scientists, robots are less attached to previous results, less driven by bias—and, crucially, easy to replicate. They could scale up experimental research, develop unexpected theories and explore avenues that human investigators might not have considered." https://lnkd.in/eZMf2Zqs

Many of our partners ask us: Can you automate in-process quality control (QC) instruments? QC instruments unlock real-time analytics and allow a robotic system to make dynamic decisions about how to continue a process based on biological factors that these instruments can measure. That is why we at Multiply Labs have automated the NC200, one of the most commonly used cell counters in GMP environments. QC instruments tend to be harder to automate because they are open systems, but our flexible technology at Multiply Labs allowed us to both automate it and make it a closed system. No other automation technology in the market can do this! Check out our video below featuring one of our brilliant senior scientists Sudeshna Sadhu to see a side-by-side comparison between a traditional manual cell count and our automated robotic cell count system. #Biotech #Automation #Robotics #Bioprocessing

Conceivable Life Sciences, a startup in Guadalajara, Mexico, is building robots and AI models to automate crucial parts of IVF. Though they’re just in the prototype stage, the robots can suck sperm into needles and place them into eggs. The cofounders say automation will challenge an industry rife with arbitrary decision-making — and it could wind up saving patients tens of thousands of dollars. Blake Dodge reports.

Robotics + AI = a new chapter in lab automation. Dash Bio's $11M seed round is more than just a financing milestone - it is a signal that investors see enormous potential in automating standard assays like ELISA with intelligence and precision. ELISA has long been a cornerstone of biomedical R&D, but also a workflow prone to variability and manual burden. Automating this process with AI-driven robotics not only accelerates timelines but also improves reproducibility, data integrity, and regulatory confidence - three pillars of successful drug discovery. The broader takeaway? We are moving from "automating tasks" to "automating knowledge," where systems do not just handle liquid but also learn, adapt, and optimize. That shift has profound implications for labs everywhere, from startups to pharma giants. The key challenge ahead will be integrating these new robotic platforms seamlessly into diverse R&D environments while ensuring data interoperability and compliance. But the direction of travel is clear: smarter labs, faster science. What other workflows do you see as ripe for this kind of AI-driven automation? https://lnkd.in/es88vtrE #LabAutomation #LabRobotics #DrugDiscovery #Automation #LRIG