Guides / Operations
The 90-day AI short-drama operating plan
A planning framework — no platform facts to verify here.
Days 1–7: lock the world
Write the premise, art-style lock, and tone. Lock 2–4 characters with face-first detail (appearance, hair, build, outfit, voice, catchphrase). Compile the consistency fragments and generate test shots before committing to a schedule — character drift discovered at episode 20 is a rewrite, at day 3 it's a prompt tweak.
Days 8–14: pilot + cadence decision
Produce episodes 01–03 fully (beats → shots → prompts → generate → log assets → compliance check → schedule). Measure your real throughput: how many finished episodes per week can you sustain on your worst week? That number, not your best week, sets the public cadence.
Days 15–45: the grind, batched
Batch production: one session for shot prompts, one for generation, one for editing/upload prep. Keep the streak — the calendar's catch-up reschedule exists for the week you get sick. Run a compliance check per platform before each first publish of an episode. Record a metrics snapshot every Monday.
Days 46–75: double down on what the numbers say
The threshold tracker now has 4–6 snapshots — enough for a linear growth estimate per metric. If the Shorts path (10M views/90d) is outpacing the watch-hours path, weight output toward Shorts. Test hooks: the first 2 seconds decide everything; rewrite hooks in the beat outline, not in the edit.
Days 76–90: apply
If thresholds are met, apply for YPP / Creator Rewards — the tracker shows "Met — apply now". If not, you have an honest ETA instead of a vibe. Either way you now operate a studio: consistent characters, a cadence, a compliance trail, and numbers.