AI Playbook 4 of 5

How to Create Presentations and Visual Content with AI

Presentations demand a different kind of AI collaboration than written documents. Slide content must be concise, visually oriented, and structured around a narrative arc that guides the audience from opening to conclusion. This section covers how to use AI for presentation development while maintaining the design judgment and data accuracy that distinguish strong decks from slide dumps.

This playbook covers the how. For the why and what, see the skill definition .
Developing Start here. Build the foundation.
  • Start every presentation project with narrative development, not slide content. Before generating any text, prompt AI: 'I am presenting [topic] to [audience]. The one thing they should remember is [key message]. What are three supporting points that build to that conclusion, and what is the strongest opening hook?' Lock the story arc before you write a single slide. This prevents the most common AI presentation failure: generating dense text for individual slides without a coherent narrative connecting them.
  • Direct AI to produce slide copy in headline-and-supporting-point format rather than paragraphs. For each slide, prompt: 'Write a headline of 8 words or fewer that captures [this point]. Below it, write 2-3 bullet points of no more than 12 words each that support the headline.' If the AI produces paragraphs or dense bullets, send them back with: 'Condense each bullet to a single line that a reader can absorb in 3 seconds.' Presentation text must be scannable, not readable in the traditional sense.
  • When using AI to summarize data for presentations, always verify the summary against your source data before including it in any slide. Prompt AI with the raw data and ask for 'the three most important takeaways for [audience].' Then check each takeaway against the original data to confirm accuracy. Keep the source data accessible so you can answer questions during the presentation. AI data summaries sound authoritative but may misrepresent the underlying numbers.
Proficient Build consistency and rhythm.
  • Use AI to generate multiple narrative structures for the same presentation. Prompt: 'Give me three different ways to structure a presentation about [topic] for [audience]. One should lead with the problem, one should lead with data, and one should lead with a story.' Evaluate each structure against your audience's needs and attention patterns. The first structure AI suggests is rarely the strongest, and spending five minutes comparing options often improves the entire presentation.
  • Apply AI image and design tools selectively based on whether they strengthen or weaken your message. Use AI-generated visuals for conceptual illustrations, placeholder mockups, and brainstorming. Avoid them for data visualizations where precision matters, for images representing real people or situations, and for any context where an audience might question whether the visual is real. Before adding any AI-generated image, ask: 'Does this serve the message, or am I including it because it was easy to create?'
  • Build a presentation template prompt library for the 3-4 presentation types you create most often. For each type, document the narrative structure, slide format requirements, audience expectations, and your most effective prompts. Reuse and refine these templates each time rather than starting from scratch. Update the library quarterly based on which approaches consistently produce strong decks.
Mastered Operate at the highest level.
  • Integrate AI-generated text, data summaries, and visual elements into a cohesive presentation using your own design judgment as the final layer. After assembling all AI-generated components, review the deck as a unified whole: does the visual language feel consistent? Do transitions between sections flow naturally? Does every slide advance the narrative? Remove or redesign any slide that feels like a disconnected AI output rather than part of an intentional story. The final design pass is always human.
  • Develop a presentation review process that separates content review from design review. First, read the deck as a text-only outline to check narrative logic and argument strength. Then review the visual design for consistency, readability, and emphasis. This two-pass approach prevents design polish from masking weak content, which is a common failure mode when AI produces visually clean but substantively thin slides.
  • After delivering a presentation, conduct a 10-minute self-debrief: which slides landed with the audience, which fell flat, and which generated questions? Map these observations back to your AI collaboration process. Were the weak slides ones where you accepted AI defaults without editing? Were the strong ones where you added your own data or restructured the narrative? Use these insights to improve your next presentation workflow.
Back to AI Content Creation Playbook

Unlock Skill Progression

Coaching Personalized to your current level
Progress Tracking Across every skill area
Mastery Validation Evidence-based, not guesswork