Set up your client project: channel voice, product claims, and past scripts
A Juma Project is a shared space where the team keeps everything Juma needs to know about a channel: how the host talks, what the product can claim, and how the videos are shot. Build one project per client or channel and every script pulls from the same context, so the tenth video sounds as on-brand as the first. Add these items once and the script stops guessing at the voice.
What to add
Channel Voice & Past Scripts
Transcripts of a few past videos, plus notes on how the host talks: pacing, running jokes, the words they always use. Drop in two or three and the script comes back sounding like the host from the first draft, not a generic narrator.
Product & Claims Notes
What the product actually does, the claims that are approved, and the words the brand will not say on camera. This keeps the demo beats and the CTA accurate, so nothing in the script needs a legal rewrite after it's shot.
Audience Profile
Who watches the channel and what they already know. This pitches the script to the right level, so a beginner explainer never talks over the viewer's head or spends a minute explaining something they already understand.
Visual & B-roll Guidelines
The channel's shot style and the footage the team already has. With this in the project, the B-roll notes point at things the team can actually film instead of describing shots that would need a second shoot day.
Guide Juma with project info
Add a short description to each knowledge item in the project's info field so Juma knows what each file is and when to use it. For example:
- Channel Voice & Past Scripts: "Transcripts of our best-performing videos. Match this voice and pacing in every script."
- Product & Claims Notes: "Approved product claims and words we never use on camera. Check before writing the CTA and the demo beats."
- Audience Profile: "Who watches this channel and what they already know. Pitch the script to this level."
- Visual & B-roll Guidelines: "Our shot style and the footage we already have. Point B-roll notes at what we can actually film."
Turn a topic into a ready-to-shoot script
Frequently Asked Questions
How much time does this Flow save compared to writing a YouTube script by hand?
Writing a six-minute YouTube script by hand takes two to four hours: shaping the hook, sequencing the story beats, adding B-roll notes, and timing the whole thing so it lands. Juma researches the brand and returns a ready-to-shoot script in a couple of minutes, so the team spends its time editing and performing rather than staring at a blank page.
The draft is a starting point, not the final cut. Strategy, taste, and judgment stay human: the team rewrites the lines that do not sound like the host, cuts the beat that runs long, and sharpens the hook before anyone picks up a camera. Human review sits on every script before it shoots.
What does the finished YouTube script include?
The script comes back ready to shoot: a timed hook, five story beats, a recap checklist, and a timed CTA with end-card suggestions. Each beat carries the on-camera lines, the B-roll notes for what to film, and a lower-third caption. Runtime math keeps it on time, and production notes flag where an emotional-payoff montage carries the middle.
Because the script is timed beat by beat, the team knows the six minutes are real before the shoot, not a guess that falls apart in the edit. The production notes call out the two or three moments that carry the video, so shoot day prioritizes the footage that matters instead of spreading effort evenly across every second.
Does the script work for a solo presenter, two hosts, or a voiceover?
Yes. Tell Juma who is on camera and the script matches the format. A solo presenter gets first-person lines to deliver, two hosts get a back-and-forth with each part attributed, and a voiceover-over-B-roll video gets narration paired with the shot list it plays over.
The format changes how the whole script reads, not just the labels. A two-host script builds in the handoffs and the moments where one host reacts to the other. A voiceover script leans harder on the B-roll notes, because the footage is doing the work the presenter would otherwise do on camera. Name the setup at the start and every beat is written the way it will actually be delivered.
Can Juma match our channel's voice?
Yes. Add a transcript of a past video or a few notes on how the host talks to the client's Project, and the script comes back in that voice instead of a generic narrator. The channel's phrasing, pacing, and running jokes carry into every new script the team runs.
For agencies running several channels, this is where one Project per client pays off. Juma remembers each channel's voice, product claims, and visual style without a re-brief, so the script for one client never bleeds into another. Unlimited seats mean the whole team, from the strategist to the editor, works from the same context on every video.
How is this different from an AI that just writes a title or a video description?
A YouTube title or description tool writes the packaging around a video that already exists. This Flow writes the video itself: the spoken script, the shot list, and the on-camera lines. It's the difference between naming the thing and building it.
The two work together. Once the script is written, the team can run Juma's YouTube title and description Flows on it to package the finished video for search and for the click. The script sets the story, and the packaging Flows dress it for the feed. Together they cover the whole path from a topic to a video that gets found and watched.