Answers are grounded in this paper and cite their sources.
The increasing use of generative artificial intelligence (AI) for long-term research, programming, academic writing, professional activities, and personal knowledge management has resulted in conversations that can accumulate thousands of messages and substantial quantities of multimodal data. As conversation histories grow, users may encounter situations in which an existing conversation becomes extremely difficult or even impossible to open, particularly on smartphones and other devices with limited memory, processing resources, or network capabilities. Conventional conversation-management mechanisms generally assume that users can first open an existing conversation and then perform operations on its internal messages. This assumption creates a fundamental usability problem: when the conversation itself cannot be opened successfully, the user may also become unable to access the conventional operations required to reorganize or reduce its size. This paper proposes Pre-Open Manual Conversation Decomposition for Generative AI (PMCD-GAI), a framework that enables users to manually divide an existing large conversation into multiple smaller, independent conversations directly from the conversation history or conversation-list interface, without opening or rendering the original conversation. Unlike conventional message-level splitting, which requires users to enter an existing conversation and select individual messages, PMCD-GAI introduces a pre-open decomposition mechanism in which the user selects the target conversation from the conversation list and specifies how it should be divided through lightweight controls. The system performs the decomposition using stored conversation metadata, message indexes, chronological identifiers, and backend data-management operations rather than requiring the complete original conversation to be loaded into the user interface. A defining feature of the proposed framework is user-controlled original-conversation lifecycle management. Users can choose to retain the original conversation after decomposition or delete it after successful creation and integrity verification of all resulting conversations. The framework supports manually specified partition counts, message ranges, chronological boundaries, and size-oriented partition configurations, while preserving message order, attachment relationships, conversation identifiers, and necessary historical references. Its central contribution is the ability to reorganize an existing conversation even when the original conversation cannot be opened through the conventional interface. This paper establishes the conceptual foundations, system architecture, formal model, operational procedures, integrity requirements, and evaluation methodology of PMCD-GAI. To the best of our knowledge, it proposes one of the first explicitly defined frameworks for user-controlled, pre-open manual decomposition of existing generative AI conversations, in which the original conversation does not need to be opened or rendered before the decomposition operation can be initiated and completed.
No takes yet. Share an insight, caveat, or question.
Jincheng Zhang (2026) studied this question.
Synapse has enriched 8 closely related papers on similar clinical questions. Consider them for comparative context: