How to Create Presentations with AI (the right way)

Jeff -

Hey friends, I get pretty annoyed when I see clickbait titles like "This AI tool completely replaces PowerPoint!"

So today, I'll show you exactly how to combine ChatGPT's Deep Research capabilities with Gamma's design features to create presentations that uncover meaningful insights AND look professional.

You'll learn a proven 3-step process that transforms raw market research into compelling business presentations, complete with tactical recommendations and data-driven storytelling frameworks used by top consulting firms.

Watch it in action

Resources

  • Jeff's Deep Research Prompt Template (see below)
  • SCQA Framework Prompt Template (see below)
  • Original Reddit post with Deep Research Prompt Generator

The Real Problem with AI Presentation Tools

Here's what most people get wrong about AI presentation tools: they're fantastic at creating beautiful slides for what I call "low-stakes presentations," but they completely fail at the most critical part of any business presentation - uncovering the insights that actually matter.

Think about it this way: there's a fundamental gap between insight generation and insight presentation. These tools can't tell you why your client should enter the US market, but they can present that insight compellingly once you have it.

The 3-Step Process That Actually Works

After years of creating pitch decks as a salesperson at Google, I've developed a three-step process that leverages AI tools for what they do best:

  1. Research Phase: Use Google Gemini's or ChatGPT's Deep Research feature to identify opportunities and risks
  2. Structure Phase: Transform findings into a structured outline that tells a clear story using consulting frameworks
  3. Design Phase: Drop that outline into Gamma for polished slides, then refine using your expertise

Let me walk you through exactly how this works using a real example: creating a market entry presentation for a budget skincare brand.

Step 1: The Research Phase

Open ChatGPT and select the latest reasoning model (the "o" models). The model versions change quickly, so just pick the newest reasoning model and toggle on "Deep Research."

Here's the critical part: Unlike normal prompts, Deep Research needs incredibly specific and comprehensive prompts. For our skincare brand example targeting young adult males aged 25-35, we're analyzing three markets: Japan, Indonesia, and the UK.

My Market Entry Prompt

ROLE Act as a senior international market-entry consultant with over 20 years of experience helping Chinese brands export overseas.

OBJECTIVE

Produce a rigorous, source-cited market-entry report for a budget skincare brand aimed at young adult males age 25-35 across these markets:

  1. Japan
  2. Indonesia
  3. United Kingdom

WORKFLOW (think step-by-step, but show only the final report)

  1. Frame the research scope and clarify any missing inputs.
  2. For each market, use Porter’s 5 Forces framework to analyze competitive forces.
  3. Weigh opportunities vs. risks, rank markets, and craft entry recommendations.
  4. Cite every external data point in IEEE style (e.g., [1]) and list sources at the end.

OUTPUT FORMAT (inside <MARKET_REPORT> tags)

<MARKET_REPORT>

  1. Executive Summary (≤ 200 words)
  2. Market Snapshots (table)
  3. Deep-Dive Factor Analysis (sub-sections per market)
  4. Opportunities & Challenges
  5. Recommendations & Next Steps
  6. Ranked Market Scorecard
  7. Reference List

</MARKET_REPORT>

CONSTRAINTS

  • Use clear, jargon-free language.
  • Back every numeric claim with a citation.
  • If data is unavailable, state “Data gap - further research required”.
  • Answer in ~800-1200 words total.

What Makes a Good Deep Research Prompt

Your prompt should include:

  • Specific target demographics and markets
  • Clear analytical frameworks (we use Porter's 5 Forces)
  • Explicit deliverable requirements (executive summary, deep dive analysis, ranked market scorecard)
  • A request for ChatGPT to clarify any missing inputs

Remember, we're not just performing market research - we're building toward a specific recommendation for our client.

The Results That Matter

After running for about 7 minutes, ChatGPT delivers:

  • A comparison table with actual market numbers
  • Deep dive analysis for each country
  • A ranked market scorecard (Indonesia ranks first)
  • Specific, actionable insights

For Indonesia specifically, the report recommends:

  • Positioning products as "formulated for tropical climates and oily skin"
  • Getting Halal certification to increase trust
  • Prioritizing marketplaces like Shopee and Lazada

This is the kind of specific, actionable intelligence that generic presentation tools simply can't provide.

Step 2: The Structure Phase

Now we have a detailed report, but we need a presentation, not a research paper. This is where strategic structuring comes in.

Back in ChatGPT (with Deep Research de-selected), we use a new prompt to create a structured outline. The key elements:

  • Target audience clarity (Chief Marketing Officer of a budget skincare company)
  • Clear recommendation (expand into Indonesia)
  • Explanation of why other markets aren't suitable
  • SCQA framework (Situation, Complication, Question, Answer) from top consulting firms
  • Pyramid Principle for each slide (key takeaway first)

My SCQA Prompt Template

Assume the role of senior business analyst with over 20 years of experience in the skincare industry.

I’m preparing a presentation for the Chief Marketing Officer of a budget skincare company based out of China looking to expand their business into new markets.

Based on the comprehensive market research I’ve already done, your task is to prepare a presentation recommending the CMO expand into the Indonesia market. Dedicate one slide to clearly and concisely explain why the other two markets aren’t suitable.

Use the Situation, Complication, Question, Answer (SCQA) framework from top consulting firms to craft a structured outline for a 8-slide presentation.

  • Situation: Functions as a starting point and a common basis. Therefore it primarily contains recognizable and agreed points.
  • Complication: Spells the reason for acting now. It contains threats / opportunities and the hurdles that need to be overcome.
  • Question: Asks the question how the hurdles of the Complication can be overcome. How can we prevent the threat or seize the opportunity?
  • Answer: Provides the answer on how to overcome the hurdles. Explains how this will help deflect the threats or seize the opportunities.

Each slide should follow the Pyramid Principle and include 3 details supporting the slide title in bullet point format:

  • Lead with the answer first: The most important part of the Pyramid Principle is to make your recommendation(s) right away.
  • Support the answer with key facts and insights at a high level: Synthesize the main takeaways from your analysis that support or even challenge your recommendations.
  • Back key insights up with detailed data, analysis, and evidence: These back-up pages will probably include many of the analyses you perform first and summaries of the evidence you gather. These data points help you and your listener understand the situation clearly and see how it all adds up to your recommendations.

Why Structure Matters

Step 1 gave us the facts. Step 2 packages those facts into a compelling story that convinces the CMO to take action. The output follows proven consulting methodologies:

  • First slide: "Enter Indonesia for highest ROI"
  • Supporting data: $9 billion opportunity, 8% annual growth
  • Contrast with slower growth: Japan (4%) and UK (2%)
  • Tactical recommendations: "search volume for 'acne wash pria' +54% YoY"

Each slide builds on the previous one, creating a logical flow from problem to solution.

Step 3: The Design Phase

With our structured outline ready, we head to Gamma. After setting up an account (the free plan works perfectly), the process is straightforward:

  1. Click Create New → Paste in text
  2. Select Presentation format
  3. Configure your preferences:
    • Text content: Choose "Condense" since we have a strong outline
    • Max text per card: Select "Detailed" (easier to delete than add)
    • AI Images: Illustrations work better than AI-generated images
    • Theme: Professional options like "Icebreaker"

The 80/20 Rule of AI Presentations

Gamma generates a beautiful presentation in about a minute, but it's only 80% complete. During your first review, check for:

  1. Flow continuity: Does the presentation still tell a coherent story?
  2. Data integrity: Did we lose any crucial information in the generation?

Common edits include:

  • Clarifying terminology ("skincare market" not just "market")
  • Using AI to improve titles ("Capturing Indonesia's 54% YoY growth: our 3-step strategy")
  • Adjusting layouts for better visual balance
  • Adding relevant stock photos or icons
  • Creating comparison slides with specific market data

Advanced Features Worth Using

Gamma includes several powerful features:

  • Spotlight mode: Blurs everything except what you click during presentations
  • Edit with AI: Quickly improves content without starting over
  • Flexible layouts: Easy drag-and-drop for images and icons
  • Quick commands: Use "/" to add buttons, links, or other elements

The Human Element: Why You Still Matter

As Professor Ethan Mollick from Wharton points out, the greatest value comes not from surrendering control entirely to AI or clinging to entirely human workflows, but from finding the right points of collaboration for each specific task.

In this workflow:

  • Humans define the business problem and objectives
  • AI researches comprehensive market data
  • Humans select the most relevant insights
  • AI structures the narrative flow
  • Humans refine the story and messaging
  • AI designs the visual presentation
  • Humans polish the final details

Making This Work for Your Presentations

This process works for any business presentation where you need both substance and style:

  1. Start with Deep Research for any topic requiring data analysis
  2. Use proven frameworks (SCQA, Pyramid Principle) for structure
  3. Let Gamma handle the design heavy lifting
  4. Always review and refine with your expertise

Remember: AI presentation tools like Gamma are incredible time-savers, but they're the visualization layer, not insight generators. Match the right tool to each phase of your process, and you'll create presentations that both inform and impress.

If you enjoyed this

Check out my full AI playlist on YouTube!