This weekly update highlights major advances in agentic AI – artificial intelligence systems that can think and act on their own. Companies like OpenAI, Google, and Nvidia are building AI agents that work like smart helpers, making discoveries and solving problems faster than ever before.

The biggest news is that OpenAI's GPT-5 can now design new science experiments by itself. In one amazing example, it helped scientists make a gene-cloning process 79 times faster. Meanwhile, a Chinese company called DeepSeek released an open-source AI model that rivals OpenAI's technology but costs much less. This is important because it means more people around the world can use powerful AI.

Google announced their new Gemini 3 and Gemini 3 Flash models that can help with many tasks, from coding software to predicting weather. Nvidia created NitroGen, an AI agent that learned to play over 1,000 video games and performs 52% better on new games than regular AI trained from scratch.

But here's something interesting: scientists discovered that when humans and AI work together, they get better results than AI working alone. A team from Stanford University and Carnegie Mellon University found that hybrid teams (humans plus AI) create much higher quality work.

Companies like OpenAI, Anthropic, and Google have formed the Agentic AI Foundation to create standard rules for how AI agents should work together. This helps different AI systems communicate better, just like how different phones can send texts to each other.

These breakthroughs show AI is becoming more useful as a partner rather than just a tool, helping scientists, doctors, and workers solve tough problems faster.