The AI agent Claude Opus 4.7 reprogrammed a robot dog 20 times faster than humans: a new frontier for physical AI
We are witnessing a landmark shift in the evolution of artificial intelligence: Anthropic's Claude Opus 4.7 neural network, during the updated Project Fetch experiment, demonstrated the ability to fully autonomously configure and control a robotic dog. The results are impressive — the AI completed tasks 20 times faster than a team of human engineers working with previous versions of the model.
While in August 2024 the AI only acted as an assistant for people without experience in robotics, Claude Opus 4.7 now operated with virtually no human involvement. Under minimal researcher supervision, the neural network independently completed the full cycle of work:
- connected to video sensors and LiDAR;
- wrote a program for manual robot control;
- created a trajectory monitoring system;
- configured an object recognition algorithm.
The key metric is speed. The Opus 4.7 model proved to be 18 times faster than the group using older AI versions and 37 times faster than people working without chatbot assistance. Moreover, the code generated by the neural network turned out to be 10 times more compact than human-written code, indicating higher algorithm efficiency.
It is important to emphasize: Anthropic did not implement specialized algorithms for controlling the hardware. The progress in robotics was a side effect of the general scaling of language models. This confirms the thesis that fundamental improvements in LLM architecture automatically expand their applicability in the physical world.
However, there were limitations. Claude struggles with precise physical manipulations: the model successfully guided the robot to the target but could not neatly push a ball to the desired point. Tasks requiring complex real-time feedback remain the prerogative of humans for now.
Anthropic believes the industry is entering an era of "physical AI agents." I fully share this view: the ability of neural networks to work with standard equipment as effectively as with software code opens the path to automating not only digital but also real-world production processes.
Cryptalist Expert Opinion: This experiment is not just a demonstration of speed. It shows that AI is beginning to understand physical cause-and-effect relationships at a level sufficient for practical application. The next step is integrating such agents into logistics, warehouse operations, and even household robotics. Investors should take a closer look at companies developing the "LLM + robotics" combination — this could become the next big trend after generative AI.