1. Define one outcome
Write the audience, offer, platform, target duration, and call to action in five lines. Avoid mixing awareness, education, and conversion goals in one video.
2. Prepare source truth
Collect the product URL, clean packshots, brand phrases, prohibited claims, and one or two references. Better source material reduces hallucinated product details.
3. Generate directions
Use Video Agent or the current canvas workflow to produce distinct concepts: founder story, product demo, problem-solution, and social proof. Ask for different hooks, not cosmetic variations.
4. Build assets
Use AI Image for product scenes or virtual try-on, AI Video for motion, and Avatar tools only where a presenter adds clarity. Keep a consistent aspect ratio and visual language.
5. Compare before polishing
Score the opening three seconds, claim clarity, product visibility, pacing, and brand fit. Drop weak concepts early so credits go to promising work.
6. Add voice and captions
Check pronunciation, caption timing, disclosure requirements, and music licensing. Never assume generated copy is legally or factually safe.
7. Export and test
Create platform-specific cuts, then measure thumb-stop rate, completion, and click-through. Feed those results into the next brief.