This weekly update highlights major breakthroughs in agentic AI systems that are transforming scientific discovery and research. Scientists and companies are developing AI agents—smart computer systems that can think, plan, and solve problems on their own—to speed up important discoveries.

One major achievement comes from Cognizant's AI Lab, which built a system called MAKER that solved a massive puzzle called the Towers of Hanoi with over one million steps and zero mistakes. This was something even the smartest AI models couldn't do before. The team also pioneered new training methods using evolution strategies, which work like nature's way of improving things over time, to make larger AI systems smarter without needing as much human help.

Another exciting development involves AI co-scientists created by Google. These agentic AI systems can read scientific papers, come up with new ideas, and test them automatically. One remarkable example shows an AI system discovering a drug that could help treat a serious eye disease called dry age-related macular degeneration. The AI found this treatment idea in just days—something that took human scientists over a decade to figure out.

FutureHouse and other research organizations created special AI agents called Robin that can handle the entire scientific process on their own. These systems search for information, suggest experiments, analyze results, and improve their ideas—all without human scientists having to stop and direct them every step.

At Stanford University, researchers introduced MetaChat, a new agentic AI platform that helps engineers solve complicated problems in advanced optics design. This shows how AI agents are becoming useful tools in many different scientific fields, not just general research.

These breakthroughs suggest that agentic AI is becoming a true partner for human scientists, helping them work faster and discover new things that could improve our world.