What problem does it solve? AI features default to chat layouts — bubbles, scrollback, composer bars — even when a structured interface would serve users better, and teams lack a method to decide the right container or escape the chat pattern. ## Core Features & Use Cases - Container decision tree: A four-question flow (enumerable need, articulability, structured answer, recall need) that determines whether a feature should be GUI, conversational entry plus stage, or pure conversation. - 6-signal chat audit: Diagnoses whether a surface reads as chat via bubbles, scrollback, composer bar, turn stacking, avatars, and sent-message mechanics, with a specified replacement for each signal. - Stage design rules and recall menu: Seven rules for composing answers as generated UI on one present-tense surface, plus recall mechanisms (chapter rail, rewind scrubber, pinned artifacts, deep links) that replace transcripts. - Use Case: A voice-first portfolio agent that never sent text bubbles still read as chat; the audit found user-side bubbles, scrollback, and a bottom dock, and replacing them with a transient caption, single re-composing stage, and question-index chapter rail made it read as a composed system. ## Quick Start Ask the AI to decide whether your AI feature should be a chatbot or a composed UI surface and to run the 6-signal chat audit on your current design.
