How do you plan a YouTube video with AI, including the title, hook and sections?
by Richard Lovatt · updated 30 October 2026
I use Underway Producer: tell it what the video should be about, and it reads YouTube to see what people actually click, writes five titles rated for how likely a stranger is to click, then breaks the chosen title into sections, each telling you what to say and what to leave for the next one. You film section by section with no script, and it edits the sections together.
In short
- Five titles come with a click rating out of five, a read before filming, not a promise
- The best-rated title is shown first; you can still keep exactly what you typed instead
- Sections are planned before you film, each with what to say and what to leave for the next
- You can film in short sections one at a time or all in one take with your hook above the camera
- It can take a while to write titles and sections, so there is a wait while it works
What planning a YouTube video with AI actually means
Planning a video usually means three separate jobs: deciding what the video is about, writing a title worth clicking, and working out what to say and in what order. Most people do none of this properly. They open the camera, talk for a bit, and either never post it or spend hours afterwards trying to cut something watchable out of a ramble.
What I wanted was something that does the planning before I press record, not after. I wanted Underway's director, Bert, to ask me what I'm making, go and look at what's actually winning on YouTube for that topic, write titles rated for how likely someone is to click, and then turn whichever one I pick into sections I can film one at a time.
I asked Bert to take one of our AI-animated feature videos, in this case Motion Graphics, a Pro Edit Without the Editor, and turn it into something I could film myself, with me as the hook and the narration instead of an AI voice.
The options people try, and where they fall short
Mean to film it and never start. The idea sits in your head and nothing gets made.
Write a rough title yourself. You guess at what sounds good, with no read on what strangers would actually click.
Record one long take and hope to cut it later. Without sections planned in advance, the edit afterwards is where most videos die.
Hire or learn an editor. It works, but it is slow, it costs money, and it puts a person between you and what you actually wanted to say.
None of these give you the two things that actually decide whether a video gets watched: a title that earns the click, and a structure that keeps the point moving once someone clicks it.
Setting up the short, step by step
I told Bert, my director, that I wanted to recreate one of the AI animated feature videos as a real short, with me as the hook and narration, dotted with B-roll. Bert checked what I already had in the vault, found the matching videos, and asked which one I wanted. I picked Motion Graphics, a Pro Edit Without the Editor, and asked for a hook-driven approach with a click-worthy title and sections.
Once I confirmed the direction, Bert locked it in as a hook-driven short and stayed with me through each setup screen, prompting me to pick or change things as they appeared rather than disappearing to do it all at once.
Writing the five titles
This is the part that takes patience. Bert scans YouTube to work out what's actually getting clicked on this topic before it writes anything, so there's a wait while the five titles and an opening get written. You can watch it happen on screen rather than it just appearing.
When it finished, I had five options: Motion Graphics and B-Roll Without an Editor, Stop Paying for Motion Graphics on Your Clips, Why Your Clips Look Unedited, Shorts Are Hard Until You Add Motion Graphics, and Can You Get a Pro Edit With No Editor. Each one carries a click rating out of five, which is a read of the title before filming, not a promise of what it will actually do once it's live.
Choosing the title
I asked Bert which one it thought was best, and it recommended Motion Graphics and B-Roll Without an Editor. I went with that one.
You don't have to take the top-rated one. There is always an option to keep exactly what you typed if none of the generated titles feel right, and you can ask it to write them again if it's close but not quite there.
Planning the sections
Once the title was locked, I asked Bert to set it up in sections so it would be easy to film and understand. It writes a structure with a hook, a body made of a few sections, and a close, and tells you roughly what to cover and what to leave for the next section, so you're not repeating yourself or leaving gaps.
For this episode, the plan came back with six sections: raw take next to the finished one, why your flat take loses people, motion captions that lift one word, the title card after your hook, B-roll goes where you name a thing, and what to do with your next take. Each one is just a name at this stage; what to actually say in it comes up when you film.
Filming in sections versus one long take
Before filming, Bert asks how you want to film it: in sections, each with its own take, or all in one go. I chose sections, because it keeps each piece short and means I can retake just one part if it isn't right, rather than redoing the whole thing.
If you film it in one take instead, you say it all in one go with your opening written above the camera, and it's joined up afterwards. For longer videos, 30 minutes or more, it treats them as chapters rather than takes: one continuous run that gets split into its own YouTube chapters.
Recording the hook
The first section is just the hook, with up to forty seconds and no clock forcing you to stop early. The guidance was to say it three or four times and it picks the best one, with an example line to work from rather than a blank page.
Once I was happy, I pressed next and moved to the next section. The point of the whole setup is that each section tells you what to say and what to leave for later, so by the time you're filming there is nothing left to plan, only to say.
What it costs in time, and who this suits
The writing stages, titles and then sections, both involve a genuine wait while it reads YouTube and works out the structure. It's not instant, and you do have to sit through a loading screen more than once, but the sections themselves are built to be filmed quickly.
This suits anyone who has an idea for a video but freezes at the blank page: what to call it, what to say first, what to say next. It is less useful if you already have a script you like and just want it filmed, since the whole point is that Bert is writing the title and the structure for you, not following one you hand it.
Questions people ask
Do I have to use the title Underway recommends?
No. Bert rates five titles out of five for how likely a stranger is to click, and puts the best-rated one first, but you can pick any of the five, or keep exactly what you originally typed as your own title instead.
What if none of the five titles feel right?
You can tell Bert what's off, what's close, or what it missed, and it writes them again rather than you having to start the whole setup over.
Can I change the order of the sections once they're written?
Yes, you can ask to rename or reorder the sections before you press next and move on to filming.
What happens if I'm not happy with one section I filmed?
You retake just that section; the sections are joined in order afterwards so you don't need to refilm anything that's already working.
Does the title click rating mean the video will actually get that many clicks?
No, it's described as a read of the title before filming, not a promise. It's a judgement on the title itself, not a guarantee of performance once it's live.