Imagine a tabletop game where the pieces move themselves, reacting to your every command and decision. This isn't science fiction, but the reality being crafted by researchers exploring the fusion of Swarm User Interfaces (SUIs) and Large Language Models (LLMs). Traditionally, tabletop games with robotic elements relied on pre-programmed actions, limiting their flexibility and responsiveness. This new research introduces the "AI-Gadget Kit," a system that empowers tabletop game gadgets with the ability to understand complex instructions, strategize, and even express emotions. The secret lies in a two-agent system. A "Coordinator" agent interprets player commands and game context, while a "Controller" agent translates these high-level instructions into specific actions for the robotic gadgets. This dynamic duo allows for sophisticated interactions, like a chess piece navigating the board or a game character enacting a dramatic scene. The researchers also developed "add-on prompts" that enhance the gadgets' abilities further. These prompts enable object manipulation, symbol visualization, non-verbal communication, and scene interaction. Imagine gadgets pushing objects, forming letters, displaying excitement, or strategically navigating a game board. The possibilities are vast. From soccer-ball-shooting games to turn-based strategy combats, the AI-Gadget Kit has been tested in a variety of scenarios. The results are impressive, showing gadgets capable of complex interactions and strategic gameplay. The researchers even tested the kit with an improvisational theater application, demonstrating the gadgets' capacity for dynamic, unscripted interactions. While this technology is still in its early stages, it has the potential to transform tabletop gaming as we know it. The future of tabletop games might be one where the line between physical and digital blurs, creating a truly immersive and interactive experience.