Every AI launch arrives at the same fork. One path is a ninety-second demo: product on screen, one thing done fast, link in the caption. The other is a twenty-minute walkthrough: a creator sitting with your product, trying it on a real task, talking through what works and what does not.
Founders tend to pick based on what they would watch. That is the wrong test. The right test is what the product needs in order to be understood, and what the audience needs in order to trust it. Those two things point in different directions more often than you would think.
When ninety seconds wins
A short demo wins when the product does one thing that is obvious the moment you see it.
Type a sentence, get a video. Paste a messy spreadsheet, get a clean one. Point a camera at a menu, get the dish described. If the whole value fits in a single before-and-after, the short format shows it, the viewer gets it, and anything longer is padding.
It also wins when the audience does not know they have the problem yet. Shorts travel. They reach people who were not looking for your product and would never have clicked on a twenty-minute video about it. For that viewer, the job is a single moment of “wait, it can do that?”, and then a link.
And it wins for AI products specifically because the category is noisy. Feeds are full of tools that look identical on a landing page. A short, in a creator’s own voice, with the product actually doing the thing, cuts through in a way a feature list cannot.
The limit of the short is trust. A viewer can be impressed in ninety seconds. They are rarely convinced. If your product costs money, needs a sign-up, touches someone’s data or asks them to change how they work, a short gets you noticed and no further.
When twenty minutes wins
A long walkthrough wins when the product is only impressive once you understand it.
Agent tools, developer tools, anything with a workflow, anything where the first result is unremarkable and the tenth is where it clicks. These products die in a short, because the short shows the unremarkable first result and moves on. They come alive when a creator sits with them long enough to hit the moment that matters, and the audience watches them get there.
It also wins when the audience is sceptical, which, for AI, is most of them. Every viewer has seen a demo that was cut to hide the failures. A twenty-minute walkthrough is hard to fake. The creator tries things. Some do not work. They say so. The product recovers, or it does not, and either way the viewer has seen something honest. The honesty is what sells, not the feature.
And it wins when the buyer is a professional who needs to justify the purchase to someone else. Those viewers want to see the settings, the edge cases, the pricing page, the bit where the creator exports the result and checks it. A short cannot give them that. A walkthrough is the demo they would have asked a salesperson for, from someone they trust more than a salesperson.
The limit of the long form is reach. Fewer people start it, and fewer still finish it. The ones who do are your best prospects, but there are not many of them, and they will not find you unless something sent them.
Why you usually need both
Look at those two limits side by side. The short reaches people who do not trust it. The long form convinces people who did not find it. Each format’s weakness is the other’s strength, which is why a launch that runs only one of them tends to feel like half a campaign.
The pairing that works is simple. Shorts go out first and wide, across the platforms where your audience scrolls. They create the “wait, what is that?” moment. The walkthrough is already live, from a creator the audience has reason to believe, and it is where the curious go next. The short is the door; the long form is the room.
This is also why the same creator can rarely do both well. The people who make a great ninety seconds and the people who make a great twenty minutes are usually different people, with different audiences and different instincts. An orbit map for an AI launch typically has both kinds on it, and the brief for each is different: one thing, fast, for the short; a real task, start to finish, for the long form.
How to decide for your launch
Three questions settle it.
Can a stranger understand the value in a single moment? If yes, you need shorts, and you may not need much else to begin with. If no, you need a walkthrough, and shorts alone will underperform.
Does using the product require the viewer to trust it with something? Money, data, their work, their time. If yes, you need the long form, whatever else you run.
Is the product cheap to try? If yes, the short can carry someone straight into the product, and the product does the convincing. If no, the walkthrough has to do that job before anyone clicks.
Most AI products answer yes to the first question and yes to the second, which is why most AI launches need both. Decide which one leads, brief each format for what it is good at, and do not ask a ninety-second video to do a twenty-minute job.