Anthropic's Project Fetch tested how well their AI model, Claude, could assist researchers in programming a robot dog to fetch beach balls. Eight researchers, none of whom were robotics experts, were split into two teams: one with Claude and one without. Team Claude completed tasks significantly faster—approximately half the time of Team Claude-less—and made substantial progress toward the goal of autonomous retrieval. The experiment revealed that access to AI not only improved task performance but also influenced team dynamics. Team Claude-less expressed more confusion and negative emotions while collaborating more among themselves. In contrast, Team Claude members worked alongside Claude, leading to better morale and efficiency. The study highlighted the challenges of connecting to hardware, with Team Claude managing to do so more effectively and avoiding common pitfalls that hindered their counterparts. In three phases, the teams faced increasing complexity. In the first phase, they simply used a controller to operate the robodog. The second phase required them to connect their own computers and develop custom software, where Claude's assistance was particularly beneficial. The final phase aimed for complete autonomy in ball retrieval. The results showed that Claude's support was crucial, especially in understanding and connecting to the robot's sensors, which proved challenging for the team without AI assistance. Overall, the findings illustrate how AI can significantly enhance human performance in robotics tasks.