You can explain your product in 30 seconds, but turning it into a sales deck takes hours.
And sales decks matter — they’re how you explain what you do, why it matters, and why people should buy, even when you’re not there.
The good news? With AI and the right prompt, you can build a solid deck in under 30 minutes.
Here’s how.
You’ve probably tried tools that make slides with AI. Most of the time, the results aren’t great. Why? Because what you put in is what you get out.
Here’s the common mistake: Prompt: “Make a sales deck for my SaaS.”
And here’s what the AI gives you:
AI isn’t magic. It just works with what you give it. If you want a good sales deck, you need to think like someone who’s trying to sell — and give the AI a clear plan to follow.
Before we get into prompts, you need to understand what a sales deck actually is.
It’s not a product brochure. It’s a guided narrative that moves someone through these reactions:
Every slide should move the lead closer to the CTA.
So let’s reverse-engineer a structure that helps AI generate that.
You’ll need a tool that can turn your prompt into slides. Some tools can also read your website, find images, and even talk through the presentation like a human.
Good ones to try:
If you just need slides, any of these will work. If you want your deck to speak and interact with people, Jotform Presentation Agent is your best bet.
Most AI tools will try to “guess” the flow of your deck unless you define it. Don’t leave it to chance.
Here’s a proven 9-slide structure that converts — used in B2B, SaaS, and services:
Include this slide structure in your prompt. Don’t let the AI invent its own.
AI needs context to generate something useful. A good prompt has these ingredients:
Here’s a founder-tested prompt template you can steal:
Create a 9-slide sales presentation for \[Product Name\], a \[type of product\] that helps \[specific audience\] solve \[specific pain point\].
Use the following structure:
1. The Problem
2. What's Broken
3. Our Solution
4. How It Works
5. Benefits
6. Social Proof
7. Comparison
8. Call to Action
9. FAQ
Focus on \[1-2 differentiators\]. Use a confident but friendly tone. Visuals should be clean and modern. End with a clear CTA: \[e.g. Book a call / Start a trial\].
That prompt gives you:
Your AI-generated deck won’t be perfect — but you can polish it in 20 minutes or less.
Here’s a tactical checklist to guide your edits:
What to edit (and why):
Slide 1: Title
Slide 3: Solution
Slide 6: Proof
Visuals:
CTA Slide:
If the AI tool you’re using lets you upload stuff — do it.
You can upload things like:
This helps the AI understand your product better. The more you give it, the better and more accurate your slides will be.
Optional: What to do after you have the deck
Once your slides look good, you can do a few extra things to make them even better.
You don’t have to do all of this — but even one of these can help more people understand and trust your product.
What's the point of AI generated comments?
Why can't your target customers always find your product? - Experience sharing
The hardest part of building in public isn’t shipping.
Solid insights here. The emphasis on narrative flow and CTA clarity feels like the missing piece in most AI decks I’ve seen.
Solid breakdown. One thing I'd add: the "What's Broken" slide is where most founders lose people. They describe the problem too abstractly.
What works better is using the exact words your customers use. I pull quotes from support tickets, Reddit complaints, or sales calls and drop them verbatim. Makes the prospect think "this person actually gets it."
The 9-slide structure is solid - I've used something similar and slide 6 (proof) is always the hardest when you're early stage.
This is a solid breakdown, especially the emphasis on defining structure before prompting. One thing I’d add is that this becomes even more effective when the context around the deck is explicit upfront as well. Things like whether the deck is for early validation, a first sales conversation, or a later-stage buyer, and whether it’s meant for inbound or outbound use, can subtly change what each slide needs to do.
I really like the structure in the prompt. I had to do a deck only once and got overboard and ended up creating a way too long and complex one.
Cheers! This is great stuff
I feel you on this one, and it's a good observation that you pointed out that she structured stuff before the prompt because that's where a lot of people miss the mark. ~
They ask AI for a great deck, and don't think about what story the deck needs to tell. I think one of the work if not the best ones I have ever seen, and I have seen a lot, is where every slide justified it's place.
If every slide doesn't, for example move you from a "why care" to a "what now" it probably doesn't belong. Just that mindset will cut a deck in half and also improve the clarity. A lot of people I see make the mistake of treating the AI output as the final.
It's good for a first draft as a rough narrator, not a copywriter. The cleanup is where the actual clarity is. A way to think of it is to outline the narrative first, use AI where it makes sense to save time, and then do a real rewrite for people not for the model. I wonder how you have seen it used in real life, do people get more value from tighter first drafts or from the mental clarity that comes from having to think more.
I really like this, this is a solid breakdown, especially the emphasis on structure before prompting.
One thing I’ve noticed is that even with a good prompt, decks fall apart if the founder hasn’t actually clarified the problem and buyer in their own head first. The AI can help articulate a narrative, but it can’t fix fuzzy thinking upstream. All of us have to be clear on objectives from the get-go.
Using a fixed slide structure like this is helpful because it forces that clarity before you ever touch the tool.
This is good !
Thank you!! I needed exactly this advice. I've been creating a vision deck for a while now, but felt lost about how to turn it into a sales pitch. I'll be sure to reference this.
Artificial intelligence is everywhere now, like an organ of our body; we're just in the process of defining its purpose. Therefore, we're trying to use it in almost every task. For example, even this response was developed by AI :)
I think your 'Step 1' (defining the structure first) is where the human touch is essential. While developing Flickle (a daily movie guessing game), I realized that AI is a great worker but a bad 'architect'. If you don't guide it correctly, it can lead you astray.
Great analysis, congratulations!
sharing this with a founder friend who will def put these tips to good use. TY!