This weekly update brings exciting news about AI agents - smart computer programs that can work on their own without human help.

Scientists shared 34 new research papers that show how AI is getting much smarter. These agentic AI systems can now think through problems step by step, just like humans do. One breakthrough called SteinerSQL helps computers understand database questions better by treating them like puzzles to solve.

A company called Dataminr created new AI agents that watch the real world for dangers. These smart programs can look at huge amounts of information in seconds and tell people about emergencies or threats right away. They work together like a team, with each agent doing different jobs but sharing what they learn.

Researchers also built an AI immune system for computers. Just like our bodies fight off germs, these AI agents protect computer networks from cyber attacks. In tests, they stopped bad hackers 3.4 times faster than old methods.

The biggest change is that AI is becoming more independent. Instead of just following rules, these new autonomous agents can make their own choices and learn from mistakes. They can handle complex jobs in hospitals, banks, and factories without needing people to watch them all the time.

Scientists found that smaller, specialized AI models can work just as well as huge ones. One study showed tiny AI programs got 98% accuracy on biology tasks. This means powerful AI won't need massive computers anymore, making it cheaper and easier for everyone to use.