Tella uses a multi-modal large-language model to understand what you're saying, your tone of voice, and gestures. It also understands what's happening in your screen recording, like whether you're moving your cursor or not. It uses this understanding to determine which aspect of your video should be emphasised, for every moment in your video. If you're enthusiastically explaining a concept, with a lot of hand waving, then it'll switch to a layout that emphasises your camera. If you're deep in a demonstration of a new feature, then it'll switch to a layout that emphasises your screen. The end-result is a more dynamic video that holds your viewer's attention better.