Chapter 26: Presentations, Slides, and Visual Communication

The deck that changes a mind. The pitch that wins the funding. The briefing that gets the program approved. Presentations are where abstract thinking becomes concrete action — and where the difference between "good enough" and "compelling" can be enormous.

They're also incredibly time-consuming to create. The combination of narrative structure, visual design, content writing, and iterative refinement means that a professional-quality presentation can easily absorb 8-15 hours of work. For most professionals, that time isn't available. The result: presentations that are completed the night before, rushed in their thinking, inconsistent in their design, and less persuasive than they should be.

AI changes this equation. Not by making slides for you — AI can't understand your audience, your organizational context, or the stakes of your specific presentation. But by dramatically accelerating the parts of presentation creation where AI genuinely excels: developing narrative structure, converting that structure into slide content, writing speaker notes, suggesting visual approaches, and adapting existing presentations for different audiences.

This chapter builds a complete AI-assisted presentation workflow, from the first conversation about audience and purpose through to a presentation-ready deck. Along the way, we'll address the most common failure modes — the AI-generated deck that looks like it was made by AI, the bullet-point-heavy structure that buries the narrative, and the "data dump" that mistakes comprehensiveness for persuasiveness.

26.1 Narrative Structure First

The single most important principle in AI-assisted presentation creation is: structure before slides.

Every persuasive presentation has a narrative — a story that moves an audience from where they are to where you want them to be. If you ask AI to create slides before you've established that narrative, you'll get a visually organized set of information that may or may not persuade anyone of anything.

The discipline of establishing narrative structure first is what separates AI-accelerated presentation creation from AI-generated content dumps.

The SCQA Framework

The SCQA framework — developed by McKinsey consultant Barbara Minto and widely used in professional communication — provides a structure for building a compelling narrative backbone for any presentation.

• S (Situation): The context the audience shares. What's the current state of things?
• C (Complication): What has changed or what problem has emerged that makes the situation insufficient or requiring action?
• Q (Question): The question the audience is left asking (often implicit): "So what do we do?"
• A (Answer): Your recommendation, proposal, or key message — the answer to that question.

Everything in a strong presentation supports the Answer. Every slide either establishes the Situation, defines the Complication, sharpens the Question, or substantiates the Answer.

Prompt template: SCQA narrative development

I'm building a presentation on the following topic:

Topic: [what the presentation is about]
Audience: [who will watch this — their role, their knowledge level, what they care about]
Purpose: [what you want the audience to think, feel, or do after this presentation]
My key message (the thing I want them to walk away believing or doing): [state it]

Help me build the narrative backbone of this presentation using the SCQA framework:

1. Situation: What context does my audience already share and accept as true?
   (1-2 sentences that establish common ground — don't include anything they'd dispute)

2. Complication: What has changed, or what problem or tension exists, that creates
   the need for my message? (This is the "but" or "however" that creates urgency)

3. Question: What question does the complication leave the audience asking?
   (Often implicit but should be articulated)

4. Answer: What is my key message — the direct answer to that question?

After building the SCQA, suggest a logical sequence of 5-7 supporting points that
substantiate the Answer. These become the skeleton of my main content slides.

Audience-Centered Structure

The most common structuring error is organizing a presentation around what you know rather than what the audience needs to hear. Your audience doesn't care about the chronological development of your research — they care about what it means for them and what they should do about it.

I have the following information I need to present:

[list your key information, findings, or points — in whatever order they come to mind]

My audience: [who they are and what they care about]
My ask: [what you want them to do or decide]
Their likely objections or questions: [what they'll push back on]

Reorganize this information from the audience's perspective:
1. What is the most important thing for this audience to hear first? (Not what happened
   first — what matters most to THEM)
2. What sequence of points builds the most logical case for my ask?
3. What objections should I address proactively, and where in the narrative?
4. What can I cut — what information is "interesting to me" but not essential for
   the audience to know to accept my ask?

Create an audience-centered outline with 6-10 slides (not counting title/closing).

💡 Intuition: The test of a good presentation outline is the "so what?" check. For every section, you should be able to complete the sentence: "I'm including this because the audience needs to understand X to accept Y." If you can't complete that sentence, the section probably shouldn't be in the deck.

26.2 Slide Content Generation

With a solid narrative outline, you're ready to generate slide content. This is where AI's content generation capabilities are most directly useful — and where the discipline of "one idea per slide" becomes the organizing principle.

From Outline to Slides

I have the following presentation outline:

[paste your narrative outline with slide topics]

For each slide in the outline, create:
1. A slide title that is a clear statement, not a category label.
   Example: NOT "Market Analysis" but "The market is growing 40% annually"
   NOT "Risks" but "Three risks require attention before launch"

2. 3-4 bullet points that support the title claim. Each bullet should be a
   complete thought, not a fragment. Maximum 12 words per bullet.

3. Speaker notes: 3-5 sentences that the presenter would say to expand on
   this slide. Written in plain, natural speech.

4. A "visual suggestion" — what type of visual would best represent this
   slide's content (chart type, diagram, image direction, etc.)

Apply the "one idea per slide" rule strictly. If a slide is trying to make
two points, flag it and suggest splitting it.

The Slide Title Discipline

Slide titles are the most underused communication tool in most presentations. Most slides are titled with category labels ("Background," "Financials," "Next Steps") that tell the audience nothing about what to think.

The alternative is assertion titles — titles that make the key claim the slide supports. An audience that reads only the slide titles of a well-structured deck should understand the entire argument.

Prompt for converting category titles to assertion titles:

Here are my current slide titles:

[list your current titles]

Convert these from category labels to assertion titles. The title should state
the key claim or insight of each slide — what you want the audience to believe
after seeing this slide.

For each conversion, also tell me: if a title is genuinely a category header
(introduction, agenda, closing) rather than a content slide, leave it as-is.

Bullets vs. Prose vs. Visuals

Not every slide should have bullet points. The decision about format should follow the content type:

• Bullet points work for lists of parallel items, sequences of steps, or multiple distinct supporting facts
• Single statement (large text, no bullets) works when one insight is so important it deserves its own slide
• Visual + caption works when the data or concept is better shown than described
• Prose paragraph almost never works on a slide — prose belongs in speaker notes or documents

I have the following slide content:

[paste slide content]

For each slide, recommend the best format: bullets / single statement / visual +
caption / prose (document, not slide). For slides where bullets are recommended,
check whether each bullet is parallel and complete, or whether some should be
restructured. For slides where a visual is recommended, describe what the visual
should show.

26.3 AI Tools for Presentations

The landscape of AI-native presentation tools has expanded significantly. Understanding what each offers helps you choose the right tool for your workflow.

Gamma.app

Gamma is one of the most capable AI-native presentation tools. It generates complete presentation structures from a brief or outline, and allows real-time regeneration of individual slides. Its AI is particularly strong at: - Generating visually cohesive slide layouts from text content - Suggesting image placements and visual treatments - Creating multiple visual themes quickly - Converting document-style content into slide format

Gamma's strength is speed and visual appeal. Its weakness is that the AI-generated structures tend toward comprehensiveness over narrative focus — you'll often need to cut and restructure to create a genuinely compelling narrative.

Best for: First drafts, visual frameworks, quick-turnaround presentations where visual quality matters and time is short.

Beautiful.ai

Beautiful.ai uses an AI-powered "smart slide" system where slide layouts automatically adapt as you add or remove content. Rather than generating content, it focuses on making slide design effortless — the AI handles layout, spacing, and visual hierarchy while you provide the content.

Best for: Teams that have their content prepared and need professional visual presentation without design skills.

Canva AI (Magic Design and Magic Write)

Canva's AI features include Magic Design (generates slide templates and layouts from a brief) and Magic Write (generates text content). The broader Canva ecosystem is valuable for creating visual assets — custom graphics, charts, infographics — that can be incorporated into any presentation tool.

Best for: Teams comfortable with Canva's design environment; particularly valuable for creating custom visual assets.

PowerPoint Copilot

Microsoft Copilot in PowerPoint can generate presentations from prompts, create slides from documents, summarize existing presentations, and suggest slide redesigns. It's embedded directly in PowerPoint's interface, making it natural for organizations already using the Microsoft 365 ecosystem. Copilot can generate presentations from Word documents, which is particularly powerful if you use AI-written documents as source material.

Best for: Microsoft 365 organizations; converting Word documents and reports into presentations; teams that need deep integration with SharePoint and Teams.

Google Slides with Gemini

Google's Gemini integration in Slides provides content suggestions, image generation, and presentation structuring assistance directly within Google Slides. For organizations using Google Workspace, the integration with Google Docs is particularly useful — generating presentations from documents, importing research, and maintaining consistent formatting.

Best for: Google Workspace organizations; teams that produce source material in Google Docs.

Choosing Your Tool

The right choice depends on three factors: 1. Your organization's existing ecosystem (Microsoft 365 → Copilot; Google Workspace → Gemini; neutral → Gamma or Canva) 2. The type of output needed (highly visual/designed → Gamma or Canva; document-to-presentation → Copilot or Gemini) 3. Your design skill level (confident designer → any tool works; limited design confidence → Beautiful.ai or Gamma provide the most assistance)

26.4 Visual Strategy

AI can meaningfully assist with the visual dimension of presentations — suggesting image directions, recommending chart types, and providing specific visual prompts.

Chart and Visualization Selection

I have the following data/information I need to visualize:

[describe what you're trying to show]

Help me choose the right visualization type:
1. What chart or visualization type best communicates this?
2. How should I structure the data to make the key insight visible?
3. What annotations or callouts should I add?
4. What is the key insight the audience should take away — and how can I
   make that insight visually dominant (rather than requiring the audience
   to discover it themselves)?

If the information is better shown as a table, framework diagram, or text-based
visual (rather than a chart), recommend that instead.

AI Image Prompts for Presentations

For slides that need photographic or illustrative images, AI can help you write effective prompts for image generation tools (Midjourney, DALL-E, Firefly) or help you identify what type of stock photography to search for.

I need images for the following slides in my presentation. The presentation is for
[audience] about [topic]. The overall visual tone is [professional/modern/warm/
authoritative — describe the feeling you want].

For each slide, suggest:
1. A description of the ideal image (subject, composition, mood, color palette)
2. A prompt I can use in an AI image generator
3. Search terms for stock photography if AI generation isn't available
4. Any images that should NOT be photographic (where illustration or abstract
   visuals would serve better)

Slides needing images:
[list slides with brief content description]

The "Show Don't Tell" Principle

AI can help identify where a visual would communicate more effectively than text:

Here is my current slide deck outline:

[paste outline]

For each text-heavy slide, suggest where a visual would be more powerful than text.
Specifically, identify:
1. Data that would be clearer as a chart than as numbers in bullets
2. Processes or sequences that would be clearer as a diagram
3. Comparisons that would be clearer as a table
4. Abstract concepts that would benefit from a visual metaphor
5. Single powerful images that could replace an explanatory paragraph

For each suggestion, describe what the visual should show and why it communicates
better than the text equivalent.

26.5 Iterating Presentations

Once a first draft exists, AI is a powerful tool for adapting, refining, and creating versions for different audiences.

Executive Version Adaptation

Here is my full presentation [or outline]:

[paste presentation content]

Create an executive version:
- Maximum [X] slides (typically 5-8 for executive decks)
- Each slide must answer: "So what does this mean for the business?"
- Remove all process and methodology detail — executives need conclusions, not how you got there
- Lead with the recommendation or key finding, not the background
- Every chart must have a "headline" that states the key insight (not just labels)
- Include only the slides that require executive attention, input, or approval

Identify which slides to cut entirely, which to merge, and which content to move
to an appendix.

Audience Adaptation

I have a presentation designed for [original audience — e.g., technical team].

New audience: [new audience — e.g., potential customer, board of directors,
or sales prospect]

New audience characteristics:
- What they care about: [priorities]
- Their existing knowledge of the topic: [low/medium/high]
- What they need to decide or do: [their action]
- Their likely concerns or objections: [what they'll push back on]

Adapt the presentation for this new audience:
1. What slides need entirely new content for this audience?
2. What slides can be kept but need reframing (different emphasis, different language)?
3. What slides are irrelevant for this audience and should be cut?
4. What new slides should I add that weren't in the original?
5. Rewrite the executive summary / opening slide for this audience.

Length Reduction

My presentation is currently [X] slides. I need to reduce it to [Y] slides without
losing the key argument.

Presentation content:
[paste slide titles and key content]

Help me reduce it:
1. Identify the slides that are "must keep" because they make the core argument
2. Identify slides that support the argument but could be cut if needed
3. Identify slides that should be moved to an appendix (available if asked about)
4. For slides you recommend keeping, suggest where content from cut slides can
   be incorporated without significantly lengthening those slides

After the reduction plan, tell me: is the core argument still intact? What, if
anything, am I sacrificing by making these cuts?

🎭 Scenario Walkthrough: Elena's 40-Page Analysis to 10-Slide Deck

Elena has completed a comprehensive organizational assessment for a client — a 40-page report with survey data analysis, interview findings, benchmarking comparisons, and recommendations. The client has asked her to present the findings to the executive team in a 30-minute slot. That's 10 slides maximum.

She uses the "executive version adaptation" prompt on her report outline. The AI's most useful contribution is identifying the three things the executive team actually needs to decide, and then rebuilding the narrative entirely around those three decisions — cutting 70% of the analysis content to appendix material.

Elena's response: "The first version AI generated was still too comprehensive — it kept too much of the analysis because AI defaulted to 'complete' rather than 'decision-focused.' I had to explicitly tell it: these executives are busy people who will only read the slides I put in front of them. Force me to choose. That constraint produced the right deck."

Her final deck: 3 slides of findings (one insight each), 3 slides of recommendations (one action each), 2 slides of supporting rationale, 1 decision slide, 1 summary. Appendix: 15 slides of the analysis detail for questions.

26.6 Presentation Coaching

AI can also function as a presentation coach — reviewing your talk track, helping you anticipate questions, and preparing you for the Q&A.

Talk Track Review

Here is the talk track I plan to use for this presentation:

[paste your speaker notes or talk track]

Review my talk track as a presentation coach:
1. Transitions: Are the transitions between slides smooth and logical? Where are
   the awkward jumps?
2. Timing: Given a [X]-minute presentation with [Y] slides, am I allocating time
   appropriately? Which sections seem too long? Too short?
3. Jargon: What terms would be unclear to [audience type]? Suggest plain-language
   alternatives.
4. Stories and examples: Where am I being too abstract? Where would a specific
   example or story make the point land better?
5. Energy and variety: Where is the talk track too monotonous or too "reporting"?
   Where could a rhetorical question, a surprising statistic, or a moment of
   directness create more engagement?

Anticipating Questions

Here is my presentation:

[paste content or outline]

I'm presenting to: [audience description]
My recommendation/ask is: [what you're asking them to do or decide]

Generate the 10 most challenging questions this audience is likely to ask,
organized by:
- Factual/data challenges (questions about the evidence)
- Strategic challenges (questions about whether this is the right approach)
- Implementation challenges (questions about how this would actually work)
- Political/organizational challenges (questions reflecting hidden concerns
  about implications, implications for their team, etc.)

For each question, draft a 3-4 sentence response that directly addresses it.

Q&A Preparation

The most stressful part of any presentation is the Q&A — particularly when you don't know which questions to expect. AI can simulate the questions of specific audience roles:

I'm presenting [topic] to an audience that includes:
- [Role 1]: They care about [what they care about]
- [Role 2]: They care about [what they care about]
- [Role 3]: They care about [what they care about]

For each audience member type, generate the 3 questions they're most likely to ask
based on what they care about. For each question:
1. The question as they'd phrase it
2. What the underlying concern actually is (which may be different from what's asked)
3. A response that addresses both the surface question and the underlying concern

26.7 Common Failures in AI-Assisted Presentations

The "AI-Looking" Deck

AI-generated presentations have a characteristic look: generic stock imagery, perfectly balanced three-column layouts, overly symmetrical graphics, and phrasing that sounds like it was written by committee. The deck doesn't look bad — it looks corporate-template-default.

How to avoid it: Use AI for content, not design. Write your own slide titles (or review AI titles carefully for phrases that feel generic). Use organization-specific examples and language. Replace AI-suggested stock images with images that are specific to your context.

Bullets Over Narrative

AI defaults to bullets because bullets are the most common slide format in its training data. But bullets are often the least persuasive format — they present information without making an argument.

How to avoid it: After generating slide content, run a "bullets to narrative" pass:

Review my slide deck. For each slide that uses bullet points, tell me:
1. Is a bullet list the best format here, or would a single assertion + supporting
   visual be more compelling?
2. If keeping bullets, do they build a logical progression, or are they just a
   list of related facts?
3. Rewrite the 3 slides where the bullet format is weakest as assertion-based
   slides with a headline and visual instead.

Missing the "So What"

The most common presentation failure is presenting findings without conclusions. The audience data without the insight. The analysis without the recommendation.

Review my presentation for "so what" failures — places where I'm presenting
information without telling the audience what to do with it or why it matters.

For each slide where the "so what" is missing:
1. What is the implicit claim or conclusion I'm not stating?
2. Rewrite the slide title or add a key insight line that makes the "so what" explicit.
3. Is there any slide that could be cut because it provides interesting information
   but doesn't actually support my argument?

⚠️ Common Pitfall: The Comprehensiveness Trap

AI is trained to be helpful and comprehensive. When you ask it to build a presentation, it will generally err toward including more rather than less — covering edge cases, adding context, providing caveats. This is the opposite of what effective presentations need.

The discipline is ruthless editing: what is the minimum content required for this audience to understand and accept my key message? Everything else is noise that dilutes the signal.

26.8 Research Breakdown: What Makes Presentations Persuasive

The research on presentation effectiveness provides important context for the AI-assisted workflow this chapter builds.

Narrative structure significantly outperforms data-first presentation. Research by Chip and Dan Heath and others consistently shows that narrative-organized information (stories that create tension and resolution) is more memorable and persuasive than information-organized content. This is the evidence base for the SCQA framework — it's not just a consulting convention but a cognitively effective approach.

Visual-verbal alignment is critical. When slide content and spoken words communicate the same message simultaneously (rather than the slide showing data while the presenter says something different), comprehension and retention improve significantly. This is why speaker notes should say the same thing the slide shows, not different things.

Slide clutter reduces retention. Studies on cognitive load in presentations consistently show that slides with more than 40-50 words of text reduce audience comprehension of spoken content, as audiences shift from listening to reading. The "one idea per slide" principle is empirically grounded, not just aesthetic preference.

Assertion titles improve information retention. Research comparing descriptive titles ("Q3 Performance") with assertion titles ("Revenue growth accelerated in Q3") shows that assertion titles produce better recall of the key point. Audiences process assertion titles as the main point, which is exactly how they should be used.

Presenters significantly underestimate Q&A preparation needs. Studies of presentation effectiveness consistently identify Q&A performance as the primary factor in post-presentation credibility assessments. Presenters who are clearly unprepared for predictable questions suffer significant credibility loss, while well-prepared Q&A responses consistently elevate perceived expertise.

Summary

Effective AI-assisted presentation creation follows a discipline: narrative first, then structure, then content, then visual. AI is most powerful at generating first drafts of each stage — but the quality of those drafts depends on how specifically you've defined the narrative purpose and audience.

The most common failure modes — AI-looking decks, bullet-heavy structure, missing "so what" — are all correctable with the prompts in this chapter. They share a common cause: using AI to fill slides with content before establishing what the presentation is actually trying to accomplish.

The workflow that works: spend 15-20 minutes establishing the narrative backbone (SCQA, audience-centered outline), then let AI generate the first draft of slide content, then spend the majority of your editing time on the places where AI content generation fails — assertion titles, "so what" clarity, and the specific examples and stories that make a generic structure feel true to your situation.

Key Concepts

• SCQA framework: Situation, Complication, Question, Answer — a narrative structure for persuasive presentations
• Assertion title: A slide title that states the key claim rather than a category label
• "One idea per slide" discipline: Limiting each slide to a single key point or claim
• Cognitive load: The mental effort required to process information; clutter on slides increases cognitive load and reduces retention
• Talk track: The words the presenter says in context with the slides; should align with visual content, not replace it
• Appendix strategy: Moving supporting detail to appendix slides available for Q&A without cluttering the main deck

Next: Chapter 27 addresses business communication — email, reports, proposals, and formal documents — where AI delivers some of the most immediately measurable productivity gains.