MemSlides: A Hierarchical Memory Driven Agent Framework for Personalized Slide Generation with Multi-turn Local Revision

2026-06-15

AI Summary

Existing approaches to personalized slide generation struggle to simultaneously preserve long-term user preferences and dynamic conversational constraints across multi-turn interactions while enabling precise local editing. This work proposes a hierarchical memory-driven agent framework that decouples memory into three layers: user profile memory, working memory, and tool memory. By integrating intent-conditioned profile modeling, working memory propagation, tool memory injection, and a sliding local revision mechanism, the framework achieves, for the first time, joint optimization of zero-shot personalization and multi-turn localized editing. Experimental results demonstrate that the method significantly improves persona alignment under diverse roles and intents, enhances the accuracy of closed-loop revisions, and ensures stable transmission of conversational preferences, yielding more reliable and efficient generation performance.

Abstract

Personalized presentation generation requires more than conditioning on a current prompt or template: agents must preserve stable user preferences across tasks, retain newly introduced preferences and constraints during multi-turn revision, and carry out local edits reliably. We propose MemSlides, a hierarchical memory framework for personalized presentation agents that separates long-term memory from working memory and further divides long-term memory into user profile memory and tool memory. User profile memory stores intent-conditioned profiles for round-0 personalization, working memory carries active preferences and session constraints across revision rounds, and tool memory stores reusable execution experience for reliable localized editing. MemSlides pairs this memory design with scoped slide-local revision, so targeted updates act on the smallest affected region instead of repeatedly regenerating the full deck. In controlled experiments, user profile memory improves persona-alignment judgments on a multi-persona, multi-intent profile bank, tool-memory injection improves closed-loop modify behavior in diagnostic matched-pair settings, and qualitative cases illustrate working memory's ability to carryover preferences. Taken together, these results suggest that effective personalization in presentation authoring depends on separating persistent user profiles, session-level working memory, and reusable execution experience across generation and localized revision.

Problem

• personalized slide generation
• multi-turn revision
• user preferences
• local editing
• memory management

Innovation

• hierarchical memory
• personalized slide generation
• multi-turn revision
• local editing
• memory-driven agent