![[2602.22839] DeepPresenter: Environment-Grounded Reflection for Agentic Presentation Generation](https://arxiv.org/static/browse/0.3.4/images/arxiv-logo-fb.png){kind=logo}
[2602.22839] DeepPresenter: Environment-Grounded Reflection for Agentic Presentation Generation
Summary
DeepPresenter introduces an innovative framework for generating presentations that adapts to user needs and incorporates environmental feedback, enhancing the quality of automated presentations.
Why It Matters
As presentation generation becomes increasingly automated, frameworks like DeepPresenter are crucial for improving the adaptability and effectiveness of AI systems. This research addresses limitations in existing methods, paving the way for more intelligent and user-centered presentation tools, which can significantly impact fields like education, business, and content creation.
Key Takeaways
- DeepPresenter autonomously plans and revises presentation slides based on environmental observations.
- The framework enhances user intent adaptation and feedback-driven refinement.
- It achieves state-of-the-art performance while being cost-effective with a fine-tuned 9B model.
Computer Science > Artificial Intelligence arXiv:2602.22839 (cs) [Submitted on 26 Feb 2026] Title:DeepPresenter: Environment-Grounded Reflection for Agentic Presentation Generation Authors:Hao Zheng, Guozhao Mo, Xinru Yan, Qianhao Yuan, Wenkai Zhang, Xuanang Chen, Yaojie Lu, Hongyu Lin, Xianpei Han, Le Sun View a PDF of the paper titled DeepPresenter: Environment-Grounded Reflection for Agentic Presentation Generation, by Hao Zheng and 9 other authors View PDF HTML (experimental) Abstract:Presentation generation requires deep content research, coherent visual design, and iterative refinement based on observation. However, existing presentation agents often rely on predefined workflows and fixed templates. To address this, we present DeepPresenter, an agentic framework that adapts to diverse user intents, enables effective feedback-driven refinement, and generalizes beyond a scripted pipeline. Specifically, DeepPresenter autonomously plans, renders, and revises intermediate slide artifacts to support long-horizon refinement with environmental observations. Furthermore, rather than relying on self-reflection over internal signals (e.g., reasoning traces), our environment-grounded reflection conditions the generation process on perceptual artifact states (e.g., rendered slides), enabling the system to identify and correct presentation-specific issues during execution. Results on the evaluation set covering diverse presentation-generation scenarios show that DeepPresenter achi...