VideoCLIP uses a deep learning model called CLIP to segment videos based on natural language queries. When a user uploads a video they can enter a simple phrase describing their request and the desired length to receive a clip of what they described.