Imagine typing in "How to make a perfect omelet" and getting a concise, instructional video, complete with helpful visuals, generated on the fly. That's the promise of a new AI framework called VG-TVP (Visually Grounded Text-Video Prompting). Researchers have developed this innovative system to create multimodal procedural plans, essentially "how-to" guides that combine text and video instructions. It works by taking a text prompt, like "How to change a tire," and leveraging existing instructional videos on the topic. The system cleverly analyzes these videos, extracts key steps, and then uses AI to generate new video segments that align with the user’s specific request. These generated videos are then seamlessly integrated with text instructions, creating a coherent and easy-to-follow guide. What sets VG-TVP apart is its ability to create dynamic, action-based video instructions, unlike previous methods that focused on static images. By incorporating human-centric actions, like showing someone actually changing the tire, the system closes the cognitive gap between instruction and execution, making learning a new procedure more intuitive. To evaluate VG-TVP, the team created a new dataset called Daily-PP, containing a diverse range of everyday tasks like cooking, crafting, and home repairs. Tests showed VG-TVP significantly outperforms traditional methods, delivering more informative and accurate procedural plans, especially for tasks with subtle nuances. VG-TVP isn't just for everyday tasks. It has the potential to revolutionize fields like education, training, and even healthcare, offering a dynamic and personalized way to learn complex procedures. While the technology is still in its early stages, it highlights the exciting possibilities of using AI to make learning more accessible and engaging. Future research is focusing on how these generated instructions improve human learning outcomes and on expanding the range of tasks VG-TVP can handle. Imagine a future where you can master any skill, from cooking a complex dish to assembling furniture, simply by typing a request. VG-TVP brings us one step closer to that reality.