How to Make a Presentation 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:


