The BBC estimated that in 2018, there were 30 million PowerPoint presentations created every day1. Most of those presentations, if I’m honest, are unremarkable.
But, there are some people who turn PowerPoints into an art form. They put unfathomable amounts of time and effort into colors, fonts, images, and transitions. They get halfway to convincing you through sheer aesthetics alone.
I am not one of those people. Which is why I was pretty excited to try out AI-generated slide decks.
I researched dozens of AI presentation tools and got hands-on with ten of them. Keep reading for a look at the top pick(s), how AI slide decks work, use cases and abuse cases, and a breakdown of the testing criteria and competition.
Note: For the top picks, I’ve created two different demo presentations, one work-related and one education-related. Apart from tweaking the themes, I did not change any of the demo content.
Additionally, this review left out the 800-lb gorilla: PowerPoint. Microsoft is adding AI features to its products at an alarming rate, and I expect to do an update once Copilot has been added to Office 365 and PowerPoint.
Overall pick: Gamma
Presentations: The History of AI and Sponsoring Artificial Ignorance.
Gamma has an simple AI-guided workflow to create presentations, though it can also be used for webpages and documents. And based on only a simple starting prompt, the initial presentations were impressive.
The content made sense, especially since I could edit the outline, but the layouts were also great. In fact, the AI-generated layouts were probably the most impressive feature I saw: Gamma created a diverse set of formats, including timelines, price tables, and icons.
But no presentation is going to be perfect on the first draft, which is where the editor comes in. Gamma’s editor is excellent. It uses a block format (similar to Notion), and there are many options for different layouts and blocks. There are also blocks for third-party apps, so you can directly embed things like Figma designs, YouTube videos, or Amplitude analytics.
Gamma also had an AI assistant baked into the app, which gives you the ability to modify slides with natural language. In practice, it worked okay, but it didn’t always understand what I was aiming for. That said, it’s prominently labeled as being in beta testing.
Overall, Gamma is a great all-around presentation tool, with a lot of functionality available out of the box.
I did have a few small critiques of Gamma, relative to Tome:
It didn’t offer AI-generated images via Stable Diffusion (but they do integrate with Unsplash for easy stock photos).
It was missing some advanced features like video narration and public comments.
The final presentations looked great, but weren’t quite as sleek as Tome’s.
Gamma is free, with 400 AI credits (creating presentations and using the AI assistant require credits). It is planning on launching a Pro plan for $10/month.
Also great: Tome
Presentations: The History of AI and Sponsoring Artificial Ignorance.
Tome, for its part, designs gorgeous presentations right out of the box. The UI is dead simple to use, and created content that made sense and looked appealing. Tome’s marketing emphasizes storytelling via its presentations, and it has easy narration baked in.
Tome’s editor was also very polished and easy to use. Like Gamma, it supports AI text-editing to easily expand, summarize, or re-write content. While it can’t do fine-grained blocks, Tome can embed content and data from apps like Airtable, Framer and Miro.
One of Tome’s great AI features was a Stable Diffusion integration, which gives you the ability to add or edit fully custom images using prompts. Of course, you can also upload your own.
Overall, Tome has some strong opinions/constraints on slide layout and design, but the presentations look great from the start.
I did have a few small critiques of Tome, relative to Gamma:
The AI-generated layouts were identical.
Similarly, the overall layout and customization options were limited.
There didn’t seem to be as many third-party app integrations as Gamma.
Tome has a free plan with 500 AI credits, and a paid plan starting at $10/month ($8/month when paid annually).
Why you should trust me
I've spent the last 10 years in Silicon Valley and the last 15 years programming. While I don't consider myself a world-renowned machine learning expert, I do have some experience with machine learning. I've built and launched dozens of software products, and I have a good eye for how they work and what their limitations might be.
For this review, I spent over 50 hours researching, evaluating, and testing different AI presentation tools. I didn't receive any compensation or payment for any of them, nor am I affiliated with any of them. That said, I'm an alumnus of both Stanford and YCombinator, and it's likely that I know affiliated folks through my extended network. I've done my best to be unbiased in my review.
How AI presentations work
AI presentation designers are likely using a large language model to structure and write the content, and then use custom software to translate that structure into presentation designs.
Depending on the custom software’s presentation structure, the apps have more or less flexibility to create different layouts and designs. Some of the most interesting layouts came from Beautiful.ai, which had timelines, word clouds, standalone quotes, and more.
Whether it’s ChatGPT or something else, these apps are likely using the same underlying LLM to power their presentation format. It’s not a coincidence that both top picks had multiple identical slide titles for the same prompt. Some of the more advanced tools also used an LLM to write custom image prompts, which were then used to generate custom images for the presentation via models like Stable Diffusion.
An interesting aspect of these tools were the different pricing structures. Most cost $8-12 per month, but had some cap. In some cases, it was the number of presentations, for others it was the number of “AI credits.” AI credits are becoming more common, especially for free tiers of products. Unfortunately, that makes it hard to compare the different plans, as each product has different credit costs per AI feature.
Use cases
Each application has slightly different strengths and weaknesses, so let’s take a look at some of the ways of using AI when it comes to designing slide decks.
Prompt-based presentations. The most common use case was designing an entire presentation from a single prompt. Whether you’re doing a presentation for school or for work, the idea is to start from a simple topic. In practice, this makes sense if you're truly out of ideas or don't know where to start.
Decktopus had a unique prompt workflow: it asks for your topic, but also your goal, audience, and timing. I imagine it's using those answers to craft a longer LLM prompt in the background.
Outline/Document-based presentations. Starting from a prompt is nice, but in practice, many people already have an outline or document that they’re presenting. So rather than assuming all of the content, apps can use AI to analyze existing content to create a presentation structure. Most of the options didn't offer this feature, and the ones that did a decent job.
AI editing. Several tools had the ability to edit slide text with AI. In general, this is becoming more and more common inside of SaaS applications, and I’m guessing it will become table stakes soon for any content creation software. If you haven't seen it, it lets you summarize, expand, or re-write text with different tones.
AI image generation. Several tools also could generate custom images using AI. These almost certainly use Stable Diffusion behind the scenes, so the outputs are similar. Some tools provide structure by offering a style option (realistic, illustration, etc).
Abuse cases
But on top of the use cases, I also want to talk about abuse cases. Technology is a double-edged sword, and clearly, generative AI has the potential to cause harm. So let's acknowledge some of the trade-offs. This isn't an exhaustive list, but only a first attempt at categorizing what is now possible.
Misinformation/Bias. If you’re a student, it’s incredibly tempting to use these tools to create a presentation and call it a day. But we know ChatGPT and other LLMs can’t be 100% trusted when it comes to facts. So any factual content not written by the author needs to be fact-checked.
Similarly, there may be implicit bias when it comes to the content that’s created. We’re still learning about how LLMs work internally and how they can be safely aligned with our interests. Proofreading and thinking critically about any slide content you didn’t write yourself is key.
Data security. With many AI tools, data security is a real concern. All of these products require you to keep a copy of your text on their servers. That said, the fallback options aren’t exactly free and open: Google and Microsoft absolutely have access to your content when using Slides and PowerPoint.
But it’s also important to understand whether AI tools are using your data to train and improve their products. For most projects, that might not be a big deal, but you might have sensitive company metrics or a new product launch that you don’t want used to train an LLM.
The selection and testing process
While it was straightforward to find these tools, selecting the test list wasn’t easy. A lot of apps have a pretty landing page without much behind the curtain. At least one of the apps I ran into had no working Twitter account or About page.
From the initial list, I narrowed the list down to 10 that I wanted to actually test. I took several factors into consideration to get a diverse mix of products. Some of the factors I used to create the shortlist:
Longevity: How long has this tool been around for? Will it still be around in a year?
AI features: How does the tool incorporate AI? Does it do so in more ways than one?
Pricing: Is there a demo or free tier available? Do the paid plans seem reasonable?
Advanced features: How does collaboration work? Public sharing/analytics? Exporting to PDF or PPT?
My testing process involved creating at least two presentations: a “business” one and a “educational” one. Where possible, I used the same single-line prompts to create the presentations.
“The history of AI”
“Sponsoring Artificial Ignorance”
In addition to the simple prompt-based creation, I tried creating presentations using an existing document, though most apps didn’t support that feature. I used ChatGPT to create the outline based on the presentation title and some extra context on the goal/audience that I was aiming for.
Once I started testing, I created a comparison table for each product and ranked it along many different factors, including:
Free tier available
Price
Content quality
Slide design
Creating from prompts vs. outlines
Themes
AI images
AI text editing
Third-party content embedding
Hosted presentations
Export formats
Team management
Custom branding/theme
Analytics
The competition
Here are the different apps that I tested (in alphabetical order):
Got feedback?
Did I miss your favorite tool? Are there other use cases that you can think of? Leave a comment and let me know! And if you want to help shape the next review, vote in the poll!
https://www.bbc.com/worklife/article/20180125-our-love-hate-relationship-with-powerpoint