Embark has developed prototype tools that use semantic and visual similarity algorithms to suggest relevant game assets to creators — for example, recognizing that a bottle is semantically similar to a bucket (both containers) or that a dead tree is visually similar to a lamp post.

Magnus Nordin's 2022 research blog described "predictive context tools" as one of Embark's three main ML research tracks, aimed at "building tools that understand the context of the game environment and assist creators create." Naavik's 2024 analysis confirmed that "Embark has also demoed tools that can help creators by suggesting relevant items based on both semantic and visual similarities." These tools were being developed both for internal use and potentially for the studio's creative platform vision. The system combines natural language understanding with computer vision to bridge the gap between what a creator is thinking and what assets are available.