Imagine giving your robot simple instructions like "Fetch my screwdriver" or "Prepare for a barbeque." Researchers are exploring how to make this possible, not with complex maps, but with the power of text. A groundbreaking new method, 'Tag Maps,' leverages the vast knowledge of large language models (LLMs) combined with the simplicity of text descriptions to guide robots through spaces. Instead of relying on traditional visual maps, Tag Maps annotate locations with plain text tags like 'sofa', 'kitchen', or 'toolbox'. This approach allows LLMs, like the one powering ChatGPT, to understand and reason about the environment. When you ask your robot to fetch your screwdriver, the LLM uses its built-in knowledge to connect 'screwdriver' with related tags like 'toolbox', 'tools', or 'workshop'. It then consults the Tag Map to locate the toolbox in the real world and plans a route for the robot. This simple yet powerful idea has been tested with real robots navigating complex lab environments. Early experiments show that Tag Maps can guide robots for tasks like finding a microwave to heat up lunch or locating a paper towel to clean a spill. While still in early stages, Tag Maps open exciting new possibilities for human-robot interaction. They offer a surprisingly memory-efficient way to empower robots with spatial reasoning, allowing them to understand and act on simple, everyday instructions, just like in a sci-fi movie. Further research could overcome current limitations like false-positive recognitions, enabling even more complex and nuanced robot tasks in the future.