A short observation
The planner that turns your brief into scenes is good. It's not magic. The same model with two different briefs about the same launch will produce two different videos, and one of them will be better than the other almost every time.
The difference isn't writing skill. It's specificity. Below is what we've noticed makes the planner produce the videos that get shipped without much editing.
Lead with the noun
The first sentence of the brief should name the thing.
"We're launching Insights, a dashboard inside our analytics product that surfaces the metrics that moved most this week."
That's a brief the planner can do something with. It knows the noun (Insights), the category (dashboard), the parent (analytics product), and the value (surfaces metrics that moved). It will compose scenes around those four pillars.
Compare that to:
"We have a new feature to announce that should help customers get more value out of the product."
Same launch, no nouns. The planner will compose around generic placeholders ("Today we're excited", "a powerful new way to") because that's all the brief gave it.
Three sentences is usually right
Briefs under two sentences underspecify. The planner fills the gap with whatever the model thinks a launch video usually looks like, and you end up with templated output.
Briefs over five sentences overspecify. The planner tries to honor every detail, scenes get crowded, and the video runs long without earning the length.
Three sentences hits the sweet spot for most release videos. One sentence on what it is, one on what it does, one on who it's for. If you can't compress your launch to three sentences, you may not be ready to ship the video yet.
Name the audience, even if you think it's obvious
"For PMMs at B2B SaaS companies" is a different video than "for marketing teams" which is a different video than no audience mentioned at all. The planner adjusts scene tone, on-screen copy, and even the suggested voice when you give it a who.
If the audience is two segments, say so. "For PMMs and demand gen leaders at B2B companies." The planner can hold both in mind. It cannot infer them.
Skip the hype words
Words the planner will quietly downweight: revolutionary, game-changing, powerful, seamless, robust, unparalleled, next-generation. Not because they're wrong, but because they don't tell the system anything about your product that helps it pick scenes.
Words the planner will hold onto: numbers, specific feature names, integration names, exact verbs ("export to PDF" not "share output"). Concrete language gives the scene composer something to put on screen.
Tell it the format you want
The brief is also where you can nudge structure:
- "Open with a problem, end with a CTA to the docs."
- "Demo-heavy, less talk."
- "Internal kickoff video, save the polish."
These hints get respected. The planner doesn't have to guess whether this is an outbound LinkedIn post or a Tuesday Slack drop.
A working template
If you want a starting point, this is the one most of our better outputs come from:
"We're shipping [name], a [category] that lets [audience] [primary action]. It [secondary benefit or differentiator]. Use it for [primary use case]."
Fill in the five blanks. Hit "Generate". Edit the result, don't write it. The point of using Pitvi is that the second draft is your first action, not the fifth.
What to do with the result
Three things to look at on the first render:
- Does the first scene name your product within the first 3 seconds? If not, edit scene 1 or strengthen the noun in your brief.
- Does the middle have a demo beat? If not, add a sentence to your brief that describes what users will see when they use it.
- Does the closing scene give a next step? "Try it free" is fine. "Read the docs" is fine. "Watch the full webinar" is fine. Just have one.
Most edits after that are taste, not correctness. The planner gave you a video. Your job is to make it yours.
