Co-Pilot is an enterprise feature, designed to help organisations manage large quantities of user-generated content and surface the best and most appropriate content for a brand.

Using deep learning algorithms, Co-Pilot studies the user behaviour of content moderators on a Stack and looks for patterns around key behaviours such as publishing or disabling content, or applying Tags. Based upon these patterns, Co-Pilot builds a moderator preference profile then applies this when analysing new content which is aggregated into a user's Stack. Recommendations based upon these insights help reduce moderation time and make it easier to find quality content.

Co-Pilot leverages a range of data sources, including user behaviour insights, to identify patterns within a Stack. These patterns are used to help train recommendation models which are used to assist in the moderation process.

Driving these models are the following data sources:

Content Metadata relates to metadata provided by the relevant source networks as well as appended data such as Language and/or Sentiment, whilst from a Visual Recognition perspective, Stackla will identify key elements within the aggregated media file, and append Tags to help surface patterns.

Content Engagement Data relates to insights provided by Trackla around Impressions, Likes, Dislikes, Comments, Votes and Clicks on Tiles currently displayed on Stackla Widgets, whilst User Behaviour data provides the final piece of the puzzle for the algorithm.