DeepSlide: From Artifacts to Presentation Delivery

Ming Yang; Zhiwei Zhang; Jiahang Li; Haoseng Liu; Yuzheng Cai; Weiguo Zheng

cs.AIApr 1, 2026

DeepSlide: From Artifacts to Presentation Delivery

Authors: Ming Yang, Zhiwei Zhang, Jiahang Li, Haoseng Liu, Yuzheng Cai, Weiguo Zheng

Organizations: School of Data Science, Fudan University, Shanghai, China

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). We present DeepSlide, a human-in-the-loop multi-agent system that supports preparing the full presentation process, from requirement elicitation and time-budgeted narrative planning, to evidence-grounded slide--script generation, attention augmentation, and rehearsal support. DeepSlide integrates (i) a controllable logical-chain planner with per-node time budgets, (ii) a lightweight content-tree retriever for grounding, (iii) Markov-style sequential rendering with style inheritance, and (iv) sandboxed execution with minimal repair to ensure renderability. We further introduce a dual-scoreboard benchmark that cleanly separates static artifact quality from dynamic delivery excellence. Across 20 domains and diverse audience profiles, DeepSlide matches strong baselines on artifact quality while consistently achieving larger gains on delivery metrics, improving narrative flow, pacing precision, and slide--script synergy with clearer attention guidance.

Figures & tables

Figure 1: Main experiment on 20 domains (left is delivery scoreboard, while right is artifact scoreboard).

Figure 2: Limitations of existing approaches and the DeepSlide .

Figure 3: Overview. Stage 1 : requirement elicitation and narrative proposal; Stage 2 : logical chain editing and evidence-grounded generation; Stage 3 : interactive slide refinement and attention-oriented augmentation; Stage 4 : rehearsal and dual-scoreboard evaluation. Effects in Stage 3: Image Focus , Text to Diagram , Keynote , Data Visualization , Motion , Background , Auto Layout (Qwen3.5 [ 24 ] as example).

Figure 4: Requirement elicitation and narrative proposal (Stage 1).

Figure 5: Logical chain editing and evidence-grounded generation (Stage 2).

Figure 6: Interactive slide refinement and attention-oriented augmentation (Stage 3), where $f_{k}^{\mathrm{src}}$ and $f_{k}^{\mathrm{dst}}$ are the original and augmented slides, $\sigma_{\mathrm{deck}}$ is the Stage 1 deck style, and $\sigma_{k-1}$ is the previous-slide style.

Method	01. AI		02. ML		03. CV		04. NLP		05. Robo		06. Sec		07. SE
Method	$S_{A}$	$S_{D}$	$S_{A}$	$S_{D}$	$S_{A}$	$S_{D}$	$S_{A}$	$S_{D}$	$S_{A}$	$S_{D}$	$S_{A}$	$S_{D}$	$S_{A}$	$S_{D}$
PPTAgent	0.82	0.56	0.80	0.54	0.80	0.54	0.80	0.70	0.83	0.64	0.79	0.64	0.80	0.61
Qwen	0.73	0.61	0.76	0.61	0.75	0.62	0.76	0.59	0.76	0.60	0.73	0.59	0.77	0.60
Coze	0.78	0.50	0.80	0.52	0.81	0.52	0.81	0.49	0.81	0.54	0.74	0.45	0.68	0.41
Gamma	0.84	0.76	0.83	0.76	0.81	0.74	0.81	0.74	0.82	0.76	0.82	0.75	0.80	0.75
Manus	0.82	0.75	0.84	0.75	0.75	0.73	0.82	0.71	0.60	0.58	0.84	0.75	0.84	0.75

Table 1: Main experiment results. Detail metrics are shown in Table 7 .

Figure 7: Secondary experiment on audience-specific evaluation.

Figure 8: DeepSlide vs. Manus, varying audience {BS, MS, PhD} and duration {5, 10, 15} (Case 1).

Design Point	DeepSlide	PPTAgent	Qwen	Coze	Gamma	Manus	NotebookLM
Category I: Content planning & structuring (reduce authoring / organization burden)
Requirement elicitation	✓	✗	✗	✗	✗	✗	✗
Source indexing	✓	✓	$\circ$	✗	$\circ$	✗	$\circ$
Outline	✓	✓	✓	✓	✓	✓	✓
Nonlinear narrative	✓	✗	✗	✗	✗	✗	✗
Category II: Delivery & rehearsal assistance (reduce speaking / rehearsal burden)

Table 2: Design points across presentation systems.

Panel of Table 3 — Table 3: Impact of removing key components on the Delivery Scoreboard $S_{D}$ (Case 3).

Panel of Figure 9 — Figure 9: Varying retrieval depth.

Appendix figures & tables3 assets

Supplementary material from the paper’s appendix.

Appendix

Agent	Responsibility
Stage 1: Requirement elicitation and narrative proposal
requirements collector	elicit audience, duration, and constraints into a requirement profile.
narrative template selector	select a narrative template consistent with the requirement profile.
logical chain generator	produce a time-budgeted logical chain (or chain graph) for the talk.
Stage 2: Logical chain editing and evidence-grounded generation
logical edge generator	add sparse cross-node links for long-range coherence and bridging.

Appendix

Table 4: Stage-wise agents and their responsibilities.

System	Open Source	License	Version&Date	URL
Manus	✗	NA	Manus 1.6 Max $\mathrm{Pro}$	https://manus.im/app
Gamma	✗	NA	Gamma Pro	https://gamma.app/
PPTAgent	✓	MIT license	PPTAgent-V2	https://github.com/icip-cas/PPTAgent
Qwen	✗	NA	Qwen3-Max	https://chat.qwen.ai/
NoteBookLM	✗	NA	NoteBookLM Pro	https://notebooklm.google.com/
Coze	✗	NA	Jan 13–16, 2026	https://www.coze.com/

Appendix

Table 5: Baseline systems used for comparison.

Panel of Table 8 — Table 8: Audience-specific dual-scoreboard results (Artifact v.s. Delivery). DeepSlide rows are highlighted.

Human-in-the-Loop AI

Human-in-the-loop AI combines human feedback, oversight, and decision-making with machine learning and autonomous systems.

See all 288 papers

SlideLab: Audience-Centered Scientific Slide Generation and Evaluation
Sep 13, 2026Vidushee Vats, Karun Sharma, Yuxia WangScientific Communication Evidence-Grounded Generation
INSAIT, Sofia University “St. Kliment Ohridski”
DeepPresenter: Environment-Grounded Reflection for Agentic Presentation Generation
Feb 26, 2026Hao Zheng, Guozhao Mo, Xinru Yan +7Agentic AI LLM Agents
Chinese Information Processing Laboratory, Institute of Software, Chinese Academy of Sciences, Beijing, China · University of Chinese Academy of Sciences, Beijing, China
Narrative-Driven Paper-to-Slide Generation via ArcDeck
Apr 13, 2026Tarik Can Ozden, Sachidanand VS, Furkan Horoz +3
University of Illinois Urbana-Champaign · Middle East Technical University

Sep 13, 2026J/K move · Enter open · S save

SlideLab: Audience-Centered Scientific Slide Generation and Evaluation

Vidushee Vats, Karun Sharma, Yuxia Wang

INSAIT, Sofia University “St. Kliment Ohridski”

Scientific Communication Evidence-Grounded Generation

Open paper AskPDF arXiv

Scientific presentations are more than summaries of research papers. They need to present the work in a coherent sequence, explain the main ideas clearly, and help the audience follow the presentation. We present SlideLab, a training-free multi-agent framework for generating scientific presentations from research papers. SlideLab first plans the presentation narrative, then builds and iteratively refines a shared slide deck using agents for content planning, visual generation, layout refinement, and grounding verification. In a blind human preference study, SlideLab was preferred over both open-source and commercial systems on 77% of papers while using roughly 4 times fewer inference tokens than the strongest open-source baseline. We also introduce ConfArena, an audience-oriented evaluation framework that simulates a conference room and assesses presentations slide by slide. ConfArena matches human system rankings and detects injected presentation problems, including falsified numbers, degraded figures, dropped slides, and shuffled slide order.

Abstract

Human-in-the-Loop AI

Explore similar work

SlideLab: Audience-Centered Scientific Slide Generation and Evaluation

DeepPresenter: Environment-Grounded Reflection for Agentic Presentation Generation

Narrative-Driven Paper-to-Slide Generation via ArcDeck