Before you build: three sentences
Write these in a note and paste them into rules or project context when the product supports it: (1) Who is my viewer and what do they want this month? (2) What do I promise that competitors do not? (3) What will I never say or do on channel? These three lines prevent vague scripts and vague research.
Day one: account and safety rails
Verify email and complete any beta or approval flow. Until you are approved, some features may be read-only or hidden—this is expected.
Set your generation language and any locale preferences under User settings.
Confirm which workspace is personal versus team. Keep experimental drafts in personal space until they are ready to share.
Skim AI credits or usage surfaces if your deployment shows them—know where to look before a heavy drafting session.
Day two: one real project, end to end
Pick one publishable idea—not ten. Create a series or standalone video, write a 5-bullet outline manually or with AI, then generate only one section of the script. Edit that section until it sounds like you. This trains your presets faster than generating full scripts you never read.
Title: working title plus a one-line viewer promise.
Outline: hook, three main beats, objection handling, CTA.
Script sample: 90–120 seconds of spoken copy—enough to judge pacing.
Day three to five: research loop
Add two to four reference channels that share your audience. For each session, log: one title pattern you could ethically adapt, one pacing choice you liked, and one thing to avoid. Turn the best insight into a script brief for your next video—not a copy.
Day six to seven: review and tighten
Read your sample script aloud; fix tongue-twisters and jargon.
Update your preset with two example sentences you love from your own past videos.
Prune rules: delete any rule that did not change outcomes—fewer, sharper rules beat a wall of text.
Ongoing habits
Weekly: one hour of research, one shipped or scheduled video, one improvement to presets or outline templates. Monthly: archive dead experiments, rename series so your dashboard stays honest, and check whether your niche sentence still matches what you publish.
If you stall, shrink the scope. A finished short script with a strong hook beats a perfect long-form plan that never records.