A launch video should show what the live product can do today. If the generated clip completes a task the build cannot, publishing it turns a product gap into a credibility problem.
At 8:12 a.m., the clip was ready. The scheduled launch was close enough that every remaining minute had a job.
In the video, a customer uploaded a file, asked the product to process it, and received a finished result. The sequence looked clean. The live build could accept the file and begin the task, but it could not reliably complete the final step shown on screen.
The founder now had two options. Publish on schedule and hope the build caught up before anyone tested it, or admit that the best part of the launch video described a product that did not yet exist.
A video can carry a claim farther than the product
In 2018, Nikola published a promotional video titled “Nikola One in Motion.” The footage showed its electric semi-truck travelling along a road.
Two years later, short seller Hindenburg Research alleged that the truck had been towed to the top of a hill and filmed rolling downhill. Nikola acknowledged that the vehicle in the video was not moving under its own propulsion, while arguing that the wording referred to the truck being in motion rather than driving under its own power.
The distinction did not rescue the claim.
The US Securities and Exchange Commission later documented the episode in its case against Nikola founder Trevor Milton. According to the SEC, the video created the false impression that the Nikola One was fully functional when it was not. In 2022, a federal jury convicted Milton on fraud charges connected to false statements about Nikola’s products and technology.
The useful part of this story for an early-stage founder comes before the investigations and courtroom. It sits at the point when a team has an impressive image, an unfinished product, and the ability to publish one as evidence of the other.
Your launch is smaller. The stakes may be a handful of pilot customers, a partner in Accra, or an investor reviewing the product from Berlin. The mechanism is the same: viewers treat a product demonstration as a claim about present capability.
The real decision is what the viewer will reasonably believe
Generated video makes this harder because it can close visual gaps without asking whether the product has closed the functional ones.
A transition can remove a failed retry. A synthetic cursor can reach a screen the live workflow never reaches. A generated result can appear in seconds even though the current build requires manual work behind the scenes.
The relevant test is simple: what would a reasonable viewer believe after watching the clip?
If they would believe the product independently completes the task, someone on the team should be able to reproduce that task in the live build. Use the same input, account type, model, permissions and connection assumptions the customer will have.
This does not require every launch video to be a raw screen recording. Editing is normal. Removing pauses, enlarging text and shortening repeated steps can make a demonstration easier to follow. The line is crossed when editing changes the capability being demonstrated.
I would ask one person who did not produce the clip to write down every product claim they think it makes. Then I would test those claims against the current build. That separation matters because the person who created the video already knows which moments were compressed, generated or repaired.
The same discipline applies beyond launch clips. I wrote about it more directly in Can I Defend Every Claim in My AI-Generated Launch Video?. A claim inventory is less exciting than another edit pass, but it gives the team something concrete to approve.
Missing the schedule is cheaper than inheriting the claim
At 8:12 a.m., the founder does not need to cancel the launch automatically. The clip can be cut back to the part the product performs reliably. The incomplete step can be labelled as a prototype, a preview or a workflow still being tested, provided that description is accurate.
There are three honest versions available.
Publish a shorter demonstration that ends at the last reproducible result. Show the intended workflow as a concept and state clearly that it is not the current live build. Or move the launch until the team can run the full sequence without manual rescue.
Each option costs something. The shorter video may feel less impressive. The concept version may attract fewer sign-ups. A delay may disrupt a campaign and disappoint people waiting for the release.
Publishing the false completion creates a different cost. Sales calls become explanations. Support requests reveal the missing step. The team starts building under pressure to make yesterday’s claim true. Limited runway moves toward repairing trust instead of learning whether customers wanted the workflow in the first place.
That is where the Nikola One video returns as a useful warning. The footage travelled farther than the vehicle could under its own power. Once published, the gap became the story.
Before the scheduled post goes live, replace the generated completion with a result someone can reproduce. If that leaves twelve seconds of honest footage, launch with twelve seconds.
Comments
No comments yet.