AI presentations: how to make a good presentation with the help of artificial intelligence and which tool to choose

What an AI presentation is, and what it is not

An AI presentation is not a new format. It is an ordinary file (PowerPoint, Google Slides, PDF or a web page) whose first version was put together by an AI presentation builder based on a topic and goal, a description of the audience or a specified document such as an offer, report or memo. The tool proposes a slide outline and copy, chooses a layout and often illustrations too.

Artificial intelligence for creating presentations is good at putting material into a logical order, cutting long paragraphs down to bullet points and quickly choosing a layout. It is less good at what matters most in a business presentation: the argument that will convince a specific customer, and figures someone is accountable for. A language model (i.e. a program that predicts the next words based on an enormous amount of text) can add “growth of 34%” that never happened, and we explain why under AI hallucinations.

The simplest way to think about it is this. AI for making presentations replaces the first hour of work, i.e. the blank slide and hunting for a layout, but it does not replace thinking about what the audience should remember and do after the meeting. Anyone who types “artificial intelligence PowerPoint presentation” into a search engine usually wants exactly that: a ready file to keep working on. They won’t get a ready decision. If presentations are part of a wider plan for you, start with guide to artificial intelligence in business.

Who saves the most work with AI presentations?

The biggest winners are teams that produce lots of similar presentations. A sales rep is preparing the third offer this week for a company in the same sector, HR is putting together onboarding material for yet another group, and marketing turns a campaign report into ten slides for the board every month. In situations like these, AI takes the mechanics off people’s hands.

• A sales offer based on a Word document or meeting notes: a draft in a few minutes, then tailoring to the customer.
• Monthly report: AI turns the results table and commentary into slides, and a human checks the figures and conclusions.
• Internal training: a procedure or manual becomes a presentation with review questions.
• Conference talk: AI helps structure the narrative and shorten the text, but the substance remains the speaker’s responsibility.
• A first version of a campaign idea: a quick draft for discussion within the team before anyone brings in a designer.

Which AI presentation tool should you choose?

Your office suite decides. If everyone has Microsoft 365, the simplest route is Copilot in PowerPoint, and Microsoft’s own guidance recommends starting from a company template, because then Copilot builds slides on existing layouts. There is also ChatGPT for PowerPoint, a ChatGPT panel in PowerPoint’s sidebar; according to the OpenAI help centre it is available on all plans, with a limited allowance on the free plan and on the Go plan. That is how you make a “ChatGPT PowerPoint presentation” without copying text between windows.

Gamma and Canva (where an AI presentation is created from within the builder) are tools from a different world. They deliver an eye-catching look faster. But it is easier to lose your corporate identity in them. We describe how to choose the whole suite, not just the slide tool, in our comparison ChatGPT, Gemini or Copilot for business.

How to make a presentation with AI step by step

Creating AI presentations gives good results when the tool is given material rather than just a topic, and the sequence below works in companies regardless of which program you happen to be using.

1. STEP 01Write the goal in one sentence

   Who is watching, what should they remember, what should they do after the meeting? For example: “The client’s board should agree to a three-month pilot.” Without a sentence like this, AI will make a presentation about everything.

2. STEP 02Prepare a source document

   A memo, an offer, a report with figures. The tool should summarise your facts, not invent its own, so enter the figures yourself, with a date and source.

3. STEP 03Open the company template

   In PowerPoint and Google Slides, start from a file with the company’s colours and typefaces, and in Gamma and Canva set up a brand theme or brand kit before generating anything.

4. STEP 04Ask for the outline first, then the slides

   Approve the outline (slide titles and one idea per slide) before the AI generates the whole thing. Correcting an outline takes a minute. Correcting 15 finished slides takes half an hour.

5. STEP 05Check every figure and every name

   Compare the slides with the source document and delete anything that isn’t in the material, because this is where made-up data most often appears.

6. STEP 06Add what AI does not know

   A customer story, an example from the sector, one strong closing line. That is the difference between a presentation that sells and one that merely looks nice.

A prompt that works: how to talk to a presentation builder

Most disappointments of the “ChatGPT is no good for presentations” kind start with the prompt “make a presentation about our company”. The model doesn’t know who the audience is, how much time they have or what matters to them. So it produces generalities. A better prompt includes the audience, the goal, source material, constraints (number of slides, length of the talk) and tone.

Also ask the tool to indicate, for each slide, the passage of the document it took the content from, because that makes it much easier to catch additions that aren’t in the material. We have collected more prompt templates, including for texts and analyses, in our guide to prompts for marketing.

How much does an AI presentation cost and what drives the budget?

The tool itself is usually the smallest item. People’s time spent on corrections costs more, and in a company that cares about its image, so does preparing a good template for AI to work with. The table shows public tool prices (in dollars, excluding VAT, as at September 2026) and the factors that determine the rest of the budget.

Brand, template and data stay in human hands

AI presentations share a common flaw. They all look alike. The same layouts, the same stock photos and three columns of icons mean that when a customer looks at five offers in a week and four look as though they came out of the same generator, the one that looks like your company wins. So the first investment should be a proper template in line with brand identity, not yet another subscription.

The second issue is graphics. An image generator will create an illustration in a few seconds, but you are not always allowed to use it as you wish, and it doesn’t always match the company’s style, as we explain under AI graphics and logos for business. If presentations are an important sales tool, think about a consistent visual system built for AI; that is the job of AI branding.

When an AI presentation pays off, and when it doesn’t

Artificial intelligence for making presentations pays off where repetition is high and errors are easy to catch, but it performs worse where every word can have legal or reputational consequences, such as investor results or a crisis situation.

AI presentations, regulation and confidentiality

Since 2 August 2026, the provisions of the EU Artificial Intelligence Regulation (AI Act) on the transparency of AI-generated content have applied. For an ordinary company presentation that a human has revised and is responsible for, this usually does not mean an obligation to label it. It is different with realistic images and video that could mislead, and we describe the details in our guide to labelling AI content from 2 August 2026.

Since 2 February 2025, Article 4 of the AI Act has also been in force, under which companies using AI systems must ensure their staff understand how to use them. A short briefing for the team on what may and may not be pasted into the tool is therefore more than just good practice. We set out a programme for such training in our article on AI training for companies.

How to check whether AI presentations have made a difference

It is easy to believe that because presentations are made faster, the company earns more. These are two different things. Measure preparation time (how many hours an offer took before and after the rollout) separately from the outcome, i.e. the percentage of offers after which the customer booked another meeting.

If results have improved, check whether AI really deserves the credit: compare them with the same months of the previous year, or keep part of the team on the old way of working for a month and compare the two groups. Coinciding in time is not the same as having an impact. We write about this on our page correlation vs impact, and if presentations are part of a wider rollout, you will find a measurement plan in our article on rolling out AI in the marketing and PR department.

Five mistakes that stop an AI presentation from selling anything

Read next