top of page

How to Make a Presentation With AI

Writer: Daniel Brooks
Daniel Brooks
Sep 3
12 min read
How to Automate Report Generation With AI


You need a presentation for Friday.


You know the subject.


You've got:


A few notes


A document


Some numbers


and:


A rough idea of what you want to say.


Then you open PowerPoint.


Blank slide.


You type the title.


Change the font.


Move the title.


Add another slide.


Spend ten minutes trying to make three boxes line up.


An hour later, you've got four slides and haven't actually thought much about the presentation.


AI can remove a lot of that work.


Modern AI presentation tools can help you:


Plan the structure


Create an outline


Write slide copy


Generate slides


Suggest layouts


Create visuals


Add speaker notes


and:


Turn existing documents into presentations.


Microsoft's current Copilot functionality can create a presentation from a prompt or reference material and then continue editing individual slides through natural-language instructions. Adobe and dedicated AI presentation platforms now offer similar prompt-to-presentation workflows.


But there's a catch.


If you type:


Make me a presentation about artificial intelligence.

you'll probably get exactly what you deserve:


A generic presentation about artificial intelligence.


The tool isn't the most important part.


The brief is.


Here's how to create something you'd actually want to present.


What Is an AI Presentation Maker?


An AI presentation maker uses generative AI to help create some or all of a slide deck.


You might provide:


A topic


An outline


Notes


A report


A document


Existing slides


or:


A detailed brief.


The AI can then turn that information into a presentation structure and generate slides.


Depending on the platform, it may also handle:


Design


Images


Charts


Speaker notes


Formatting


and:


Brand styling.


For example, Microsoft's current AI presentation workflow can take a description or source material, suggest an outline, generate slides and then let you continue refining the deck.


That's very different from simply asking an AI chatbot:


Give me ten slide titles about cybersecurity.

The output can now be the actual presentation.


Can AI Make a Whole Presentation?


Yes.


But:


Can AI make a presentation?


and:


Can AI make a good presentation?


are slightly different questions.


AI can generate:


10 slides


Headings


Bullet points


Images


Layouts


and:


Speaker notes


in a remarkably short time.


What it doesn't automatically know is:


What does this presentation actually need to achieve?


Imagine you're creating a presentation about declining customer retention.


Is the purpose to:


Inform management?


Request additional budget?


Recommend changing the product?


Explain what went wrong?


Persuade the board to hire more staff?


Same subject.


Completely different presentation.


That's why your workflow shouldn't start with:


“Create slides.”


It should start with:


“What does this presentation need to accomplish?”


How Does an AI Presentation Maker Work?


The basic process is:


1. Brief


Define the presentation.


2. Source


Give AI reliable information.


3. Structure


Plan the story.


4. Generate


Create the first slides.


5. Edit


Remove weak or unnecessary content.


6. Verify


Check facts and numbers.


7. Present


Make the slides work for an audience.


Recent 2026 guides from Adobe and specialist presentation platforms recommend essentially this workflow: establish the audience and objective, provide useful source material, generate a draft, then edit and verify it rather than treating the first AI output as finished.




Step 1: Decide What the Presentation Needs to Do


Before choosing an AI tool, finish this sentence:


After this presentation, I want the audience to...

For example:


...approve a £20,000 marketing budget.

Or:


...understand why the project is four weeks behind schedule.

Or:


...choose our company instead of the other suppliers.

Or:


...understand the research findings.

Now you have a purpose.


Next define the audience.


For example:


Audience: Senior management


Purpose: Approve £20,000 marketing budget


Length: 10 minutes


Desired outcome: Budget approval


That's already a much better AI prompt than:


Make me a marketing presentation.

Step 2: Gather Your Source Material


If the presentation involves real information, give AI the information.


That might include:


Report


Meeting notes


Spreadsheet


Research


Project brief


Product information


Customer feedback


Sales figures


Existing document


This is particularly important for business presentations.


Suppose you ask:


Create a presentation explaining our Q2 sales performance.

Unless the AI has your actual Q2 sales figures, what exactly is it supposed to say?


Don't make it guess.


Provide:


Q1 revenue: £420,000


Q2 revenue: £485,000


Target: £500,000


New customers: 37


Previous quarter: 29


Largest growth area: Enterprise


Main problem: SME churn increased


Now the presentation can be built around something real.


Step 3: Write a Proper Presentation Brief


Instead of:


Make a presentation about our Q2 results.

try:


Create a 10-slide presentation for the senior management team explaining our Q2 sales performance.


The presentation should show what changed compared with Q1, explain where we missed target and finish with three recommended priorities for Q3.


The audience already understands the business, so don't include introductory slides explaining the company.


Keep slides concise and suitable for a 10-minute presentation.


Use only the data supplied below.


Do not invent missing numbers or explanations.


DATA:


[insert data]


That's a proper brief.


It gives AI:


Audience


Purpose


Length


Structure


Context


Constraints


and:


Source material.


Step 4: Generate the Outline Before the Slides


This is one of the easiest ways to improve an AI presentation.


Don't immediately generate the deck.


Ask for the structure first.


For example:


Before creating any slides, propose a 10-slide outline.


For each slide give:


  • slide title

  • one-sentence purpose

  • key evidence required


The presentation should build towards a decision rather than simply summarising the data.


Now inspect it.


Maybe AI suggests:


Slide 1


Q2 Sales Performance


Slide 2


Executive Summary


Slide 3


Revenue vs Target


Slide 4


Quarter-on-Quarter Growth


Slide 5


Customer Acquisition


Slide 6


SME Churn


Slide 7


Enterprise Growth


Slide 8


What Drove Performance


Slide 9


Q3 Priorities


Slide 10


Decision Required


That's a coherent story.


If Slide 4 repeats Slide 3, remove it.


If the important problem doesn't appear until Slide 8, move it.


It's much easier to fix:


10 slide titles


than:


10 fully designed slides.


Step 5: Give Every Slide One Job


A common presentation mistake is trying to make every slide contain everything.


Imagine a slide titled:


Q2 Performance


and underneath it:


Revenue


Customers


Marketing


Pipeline


Churn


Product updates


Support


Forecast


That's not a slide.


That's a report squeezed onto a rectangle.


Instead, give each slide one clear purpose.


For example:


Revenue Grew 15%, but We Still Missed Target


Now the audience immediately knows the point.


The chart provides the evidence.


You explain the detail.


That's stronger than:


Revenue


A useful rule is:


One slide = one main message.


Step 6: Generate the First Draft


Now generate the deck.


Different AI presentation tools handle this differently.


Some ask for:


Topic


Presentation length


Style


and:


Audience.


Others allow you to upload source material.


The AI then produces:


Slide structure


Text


Visuals


and:


Layouts.


This is the point where dedicated presentation tools can save a significant amount of formatting time.


But treat the result as:


Draft 1.


Not:


Finished presentation.





Which AI Tool Should You Use?


There are three main routes.


Option 1: Dedicated AI Presentation Maker


A tool such as Gamma is built specifically around generating presentation-style content from a prompt or source material.


This route makes sense when:


You want something quickly


You don't want to design slides manually


You're happy working outside traditional PowerPoint initially


and:


You want AI handling much of the layout.


This would be my main specialist-tool route for the article.


Option 2: AI Inside PowerPoint


If your organisation already works in PowerPoint, staying there may make more sense.


Microsoft PowerPoint now integrates Copilot directly into the presentation workflow.


You can create a presentation from a prompt, use source material, generate an outline and continue refining slides with Copilot. Availability depends on the relevant Microsoft 365/Copilot access.


The big advantage is obvious:


You're already in PowerPoint.


There's less:


Generate somewhere else → Export → Fix everything.





Option 3: AI Inside Google Slides


If your organisation works primarily in Google Workspace, Google Workspace has Gemini functionality inside Slides.


Google currently supports AI-assisted tasks including generating slides, creating images and referencing Drive content when building slides.


Again, there's an advantage to using AI inside the software your team already uses.


Don't move your entire presentation workflow to another platform just because somebody calls it:


The best AI presentation maker.


What About Canva?


Canva is another logical option when the design side matters heavily.


It's particularly useful if you're already creating:


Social graphics


Documents


Marketing materials


and:


Presentations


inside the same design system.


For some users, that's more useful than adding another dedicated presentation subscription.


Step 7: Rewrite the Slide Titles


AI-generated presentation titles are often painfully generic.


You might get:


Market Overview


Key Challenges


Results


Opportunities


Conclusion


They're technically correct.


They're also saying almost nothing.


Change:


Results


to:


Revenue Grew 15%, but Finished £15k Below Target


Change:


Customer Acquisition


to:


We Added 28% More Customers Than Q1


Change:


Key Challenge


to:


SME Churn Is Now Our Biggest Growth Constraint


Now someone could skim only the slide titles and understand the story.


That's a much stronger deck.


Step 8: Remove Most of the Text


AI loves filling space.


If the slide has room for another paragraph, AI seems determined to put one there.


Don't let it.


Your presentation isn't a Word document displayed horizontally.


If a slide contains:


Headline


Three paragraphs


Six bullets


Two charts


and:


A stock image


something has gone wrong.


Ask:


What does the audience actually need to see?


Maybe it's:


One sentence


One chart.


You can say the rest.


How Much Text Should Be on a Slide?


There's no universal word count.


A technical training slide may legitimately need more information than a keynote slide.


But use this test:


Can someone understand the slide while also listening to me?


If they're silently reading a wall of text for 45 seconds, they're no longer listening.


AI should help you reduce content as well as create it.


Try:


Reduce the text on this slide by 50% without removing the main argument or supporting evidence.

Then edit it yourself.


Step 9: Replace Generic AI Images


AI presentation generators often add visuals automatically.


Sometimes they're useful.


Sometimes you get:


Businesspeople shaking hands


People pointing at holographic charts


Futuristic glowing brains


Random laptops


and:


People smiling enthusiastically at spreadsheets.


Delete anything that doesn't help explain the point.


For a business presentation, actual:


Product screenshot


Chart


Diagram


Customer example


Photograph


or:


Process


will often be more useful.


The goal isn't:


Every slide needs a picture.


The goal is:


Every element should earn its place.


Step 10: Use AI to Improve Individual Slides


You don't need to regenerate the entire presentation because Slide 7 is bad.


Work slide by slide.


For example:


Rewrite this slide around one key message: customer acquisition improved, but the gains are being offset by SME churn.


Maximum 40 words.


Suggest one chart that would support the message.


Or:


This slide is too complicated.


Identify the single point the audience needs to remember and suggest a simpler layout.


This is where AI becomes particularly useful.


Not:


Make the presentation for me.


But:


Help me improve this specific problem.


Step 11: Use AI to Create Speaker Notes


Once the slides are good, AI can help with what you're going to say.


For example:


Create speaker notes for this slide.


The presentation is for senior management.


I have approximately 45 seconds for this slide.


Don't simply repeat the text visible on screen.


Explain what the numbers mean and why they matter.


Now the slide remains concise while the notes provide the detail.


Microsoft currently lists speaker-note generation among Copilot's PowerPoint capabilities.


Don't Read AI Speaker Notes Word for Word


They should help you prepare.


Not turn you into an audiobook.


Rewrite phrases you wouldn't naturally say.


Remove unnecessary introductions.


Add your own examples.


Practise the presentation aloud.


If you stumble over a sentence every time you say it:


Change the sentence.


How to Make AI Presentations Look Better


The biggest improvement usually isn't adding more design.


It's removing things.


Take a typical AI-generated slide:


Market Opportunity


The artificial intelligence market is experiencing significant growth driven by technological innovation, increased adoption and changing customer expectations.

Then:


Five bullets


Three icons


Stock image


Decorative shape


Instead, ask:


What's the point?


Maybe:


AI Adoption Among Our Customers Doubled in 12 Months


Then show the evidence.


Much stronger.





Use a Consistent Visual System


Your presentation doesn't need 12 different layouts.


Define:


Heading style


Body font


Accent colour


Background


Chart style


Image style


Spacing


Then repeat them.


AI sometimes treats every slide as an opportunity to demonstrate another design idea.


Don't.


A presentation should look like:


One presentation.


Not:


12 Canva templates that met for the first time this morning.





Can AI Turn a Document Into a Presentation?


Yes, and this is one of the most useful workflows.


Suppose you have a:


20-page report


and need a:


10-minute presentation.


Instead of manually copying sections, use AI to identify:


Main argument


Key findings


Important evidence


Recommendations


and:


Decision required.


Then build the presentation around those.


Microsoft currently supports creating PowerPoint presentations using referenced files, although the precise workflow varies across Copilot experiences and rollout status.


But don't tell AI:


Turn every section of this report into a slide.

A document and presentation are different formats.


The presentation needs to answer:


What does this audience need from the report?


Can AI Turn a PDF Into a PowerPoint?


Depending on the platform and workflow, yes.


But again, conversion isn't the real goal.


If you have a:


60-page PDF


you probably don't want:


60 slightly prettier slides.


You want the important information extracted and reorganised into a presentation.


Prompt around the desired outcome:


Use the attached report to create a 12-slide presentation for senior management.


Focus on findings that affect next year's budget.


Ignore background information the audience already knows.


Every recommendation must be traceable to information in the source document.


That's much better.


Can AI Create Presentations From Notes?


Yes.


This is probably the simplest workflow.


You might have:


  • Q3 project update

  • Launch delayed two weeks

  • supplier issue fixed

  • testing starts Monday

  • £8k over budget

  • need decision on contractor

  • launch now 14 Nov


AI can turn that into:


Slide 1


Project Update


Slide 2


Launch Moves to 14 November


Slide 3


Supplier Issue Has Been Resolved


Slide 4


Testing Starts Monday


Slide 5


Project Is £8k Over Budget


Slide 6


Decision Required: Additional Contractor


That's where AI saves real time.


You're providing the knowledge.


AI is helping structure it.


Can AI Make a PowerPoint Presentation?


Yes.


If PowerPoint is specifically where you need the final deck, Microsoft's own AI functionality is now one of the most direct routes.


Copilot can currently help create presentations, add slides, draft content and generate speaker notes inside PowerPoint for eligible users.


Alternatively, some dedicated presentation generators allow export to PowerPoint.


But always test the exported file.


Look for:


Font substitutions


Moved objects


Broken layouts


Image cropping


Animation issues


and:


Missing elements.


A deck that looks perfect in one application doesn't always survive export perfectly.


Can AI Make a Presentation for Free?


There are free or limited-access routes, although features and allowances change frequently.


But don't choose purely on:


Which one costs £0?


If you already pay for:




or:



check what AI presentation functionality is included or available within your existing setup first.


Adding another subscription isn't automatically the best option.


Can Students Use AI to Make Presentations?


Technically, yes.


But follow your school or university's AI policy.


There's also a practical issue.


If AI creates:


Argument


Research


Slides


and:


Speaker notes


and you don't understand any of them, the presentation itself may become uncomfortable very quickly when someone asks:


Why did you include this?

Use AI to help organise and present work y


Can AI Create a Pitch Deck?


Yes, but this is an area where generic AI output can be particularly dangerous.


An investor deck needs actual:


Market


Traction


Revenue


Customers


Economics


Competition


and:


Strategy.


AI can structure and visualise those.


It can't legitimately invent them.


A beautiful slide claiming:


£4.2bn addressable market

isn't useful if nobody knows where £4.2bn came from.


For important decks:


Every material claim needs a source.


Fact-Check Every AI Presentation


This is non-negotiable.


Check:


Statistics


Dates


Names


Quotes


Calculations


Market figures


Product claims


Sources


and:


Conclusions.


If you provided the source data yourself, verify the presentation against it.


If AI researched anything externally, open the source.


Don't trust a citation merely because it looks like one.


Look for AI's Favourite Presentation Problems


Before presenting, check for:


Repetition


Are Slides 4 and 7 making essentially the same point?


Generic language


“Driving innovation through strategic transformation.”


What does that actually mean?


Too much text


Cut it.


Weak headlines


Replace labels with conclusions.


Unsupported claims


Verify or remove them.


Irrelevant images


Delete them.


Fake precision


Where did that 37.4% come from?


Too many slides


Does Slide 9 actually need to exist?


A Simple AI Presentation Workflow


For most people, I'd use:


Step 1


Define:


Audience + Purpose + Outcome



Step 2


Gather reliable source material.



Step 3


Ask AI for an outline.



Step 4


Fix the structure manually.



Step 5


Generate the slides.



Step 6


Rewrite the headlines.



Step 7


Remove unnecessary text.



Step 8


Replace weak visuals.



Step 9


Verify every factual claim.



Step 10


Generate speaker notes.



Step 11


Practise.


That's enough.


How Much Time Can AI Save on Presentations?


It depends heavily on the deck.


If you normally spend:


30 minutes planning


60 minutes writing


90 minutes formatting


30 minutes finding visuals


then the first-draft process might consume several hours.


AI can compress a substantial amount of:


Structuring


Drafting


and:


Formatting.


But I'd use the saved time to improve:


Argument


Evidence


Visuals


and:


Delivery


rather than simply making three times as many slides.


Is It Worth Using AI for Presentations?


Yes — particularly for the first draft.


Presentations combine several tasks AI handles reasonably well:


Organising information


Summarising


Rewriting


Generating layouts


and:


Creating alternatives.


But AI still doesn't know:


Which argument matters most to the people sitting in the room.


That's your job.


The best workflow isn't:


Prompt → Presentation → Present


It's:


Brief → AI Draft → Human Judgement → Better Presentation


Start With Your Next Presentation


Don't download six AI presentation tools.


Take the next presentation you genuinely need to make.


Before opening anything, write:


Who is the audience?


What do they already know?


What do I need them to understand?


What do I want them to do afterwards?


What evidence supports that?


Then give that information to the AI.


Ask it for the outline first.


Fix the story.


Generate the slides.


And edit ruthlessly.


Because AI has made it incredibly easy to generate:


20 slides.


The valuable skill is still knowing:


Which 8 you actually need.




 
 
bottom of page