Imagine telling an AI what kind of robot you need and having it design one for you. That's the exciting promise of new research using large language models (LLMs) to design soft, modular robots. Traditionally, designing these flexible robots has been a painstaking process of trial and error, requiring significant engineering expertise. Researchers have found a way to leverage the power of LLMs, like those behind chatbots, to simplify this process dramatically. By framing robot design as a language problem, they can feed instructions to an LLM, which then generates a blueprint for the robot's structure. To test these designs without building physical prototypes, a simulation tool provides feedback, allowing the LLM to refine its design iteratively. This approach has shown remarkable success in creating robots that can walk, navigate stairs, and even move back and forth. Interestingly, the AI sometimes comes up with designs that are unconventional yet highly effective, showcasing its potential to go beyond human intuition. While promising, there are challenges to overcome. Simulating the soft, flexible materials of these robots is complex, and translating the AI's designs into real-world robots presents engineering hurdles. But this research opens up thrilling possibilities for the future of robotics, pointing toward a future where designing a custom robot might be as simple as typing out a request.