DeepSlide: From Artifacts to Presentation Delivery
DeepSlide is a new multi-agent system designed to enhance the presentation delivery process. It focuses on improving narrative flow, pacing precision, and slide-script synergy while maintaining high artifact quality. The system integrates various components to support the full presentation preparation process, from planning to rehearsal.
- ▪DeepSlide optimizes both the artifact quality and the delivery process of presentations.
- ▪The system includes features like a controllable logical-chain planner and attention augmentation.
- ▪DeepSlide has shown significant improvements in delivery metrics across diverse audience profiles.
Opening excerpt (first ~120 words) tap to expand
Computer Science > Artificial Intelligence arXiv:2605.15202 (cs) [Submitted on 1 Apr 2026] Title:DeepSlide: From Artifacts to Presentation Delivery Authors:Ming Yang, Zhiwei Zhang, Jiahang Li, Haoseng Liu, Yuzheng Cai, Weiguo Zheng View a PDF of the paper titled DeepSlide: From Artifacts to Presentation Delivery, by Ming Yang and 5 other authors View PDF HTML (experimental) Abstract:Presentations are a primary medium for scholarly communication, yet most AI slide generators optimize the artifact (a visually plausible deck) while under-optimizing the delivery process (pacing, narrative, and presentation preparation).
…
Excerpt limited to ~120 words for fair-use compliance. The full article is at arXiv cs.AI.
Discussion
0 commentsMore from arXiv cs.AI
