AI agents are becoming physical: Claude Opus 4.7 controls a robot dog 37 times faster than a human
We are witnessing a landmark shift in the development of artificial intelligence: language models are transitioning from purely digital tasks to controlling physical objects. The new phase of Anthropic's Project Fetch experiment demonstrates that AI is not just capable of assisting humans, but can fully autonomously manage complex robotics, and at a colossal speed advantage.
The Claude Opus 4.7 model completed the full cycle of configuring and programming a robot dog 20 times faster than a team of human engineers in a similar experiment a year earlier. But the most impressive comparison is with untrained users: here, the AI was 37 times faster, and the volume of code it wrote was 10 times more compact.
Autonomy at a New Level
Unlike the previous phase, where the AI acted as an assistant, Claude Opus 4.7 worked with virtually no human intervention. The neural network independently:
- connected to video sensors and LiDAR;
- wrote a program for manual control;
- developed a trajectory monitoring system;
- configured an object recognition algorithm.
Notably, Anthropic did not embed specialized algorithms for working with hardware into the model. According to the developers, the progress in robotics became a side effect of the general scaling of language models. This confirms the hypothesis that universal AI systems can adapt to physical tasks without additional training.
Boundaries of Capabilities
Despite the impressive results, Claude still lags behind humans in fine motor skills. The model successfully guided the robot to its target but could not neatly push a ball to the right spot. Such tasks require complex real-time feedback—an area where humans still hold the advantage.
Anthropic predicts that the industry is entering an era of "physical AI agents." In the near future, neural networks will be able to work with standard tools and equipment as effectively as they currently do with software code.
Expert opinion: Anthropic's breakthrough is not just another benchmark. It is a demonstration that AI agents are ceasing to be virtual assistants and are becoming full-fledged participants in the physical world. For the crypto industry, this means potential automation of mining farms, logistics for DePIN projects, and even robotic trading systems. However, the case with the ball reminds us that to achieve full autonomy in the real world, we still need to bridge the gap between "understanding" and "action."