Awdisea is an AI powered platform that helps YouTubers predict in advance how well a video idea will succeed before even producing the video. The accuracy depends on the amount of data YouTubers provide as well as analytics retrieved from various YouTube channels related to the video content that is to be produced. At the very minimum, it can make a fairly good prediction based solely on a YouTube video thumbnail and the title. The model is trained on data provided by YouTubers and uses human reinforcement feedback to correct the model after a video has been published. The platform supports A/B testing of thumbnails and titles but unlike traditional A/B testing that only support 2 versions of something, Awdisea supports up to 3 different thumbnails and 3 different titles. The AI together with human feedback will rate the thumbnails and titles and reports are sent to YouTubers to decide which thumbnail and title to use. Awdisea's AI is based on breaking down the creative process that goes on inside human minds of creative people who are able to produce creative and engaging content and model these processes along with attributes that tend to steer creative people towards coming up with successful content. The model takes into account YouTubers who create similar content rather than place all creators in a single category.

Awdisea

Predicting YouTube video views with AI during pre-production

Predicting YouTube video views with AI during pre-production