To explore entity collection and use, let's take the theoretical case of a chat assistant for a big, nameless Seattle-based coffee chain. Users often enter the app with a narrow goal, or two intents: order a coffee and check out. However, customers tend to buck the script, providing non-formulaic inputs that require flexible responses from the assistant. To handle these inputs, the assistant must manage context switching, act upon conditional logic, follow up on incomplete utterances, and preserve and recall earlier inputs for later use. The perceived intelligence of the assistant is crucial in how it captures and stores overfilled information, which can be categorized as underfilled, filled, or overfilled. Overfilling occurs when a user provides too much relevant information, while underfilling happens when they provide too little. A well-designed entity collection system is essential to handle complex intents and entities, as the complexity compounds with more entities involved. Collecting and mapping entities correctly is key to delighting users who are "overfillers" or power users, requiring thoroughness and extensive user testing.