Posting tonight, but the video still isn’t there

Last night I was staring at the publish time, blanking out: the copy was done, but I still had no visuals. I’ve been stuck here too. I used to think “one-click AI generation” would make this easy, but the part that ate the most time was rework.

What this method is, and who’s already using it

I came across a hands-on test that put Claude Fable 5 and GPT-5.6 Sol into the same $100 budget for making a music video. My takeaway was simple: the former feels more like a patient planner, helping the story, shots, and pacing make sense first; the latter feels more like a fast editing assistant, better for turning ideas into watchable clips quickly.

At 9 p.m. last Tuesday, in a coworking office in Binjiang, Hangzhou, A-Ning, who runs an indie women’s clothing brand, was editing product pages while using AI to generate a teaser video for a new release. She told me the real time savings didn’t come from “making a video.” They came from doing less back-and-forth prompt rewriting—in other words, fewer rounds of rewording the instructions sent to the AI. I’ve messed this part up before too. I always wanted to get everything right in one shot, and that ended up making me slower.

What it costs to copy this today

Money: about $100 or less is enough for one trial run. Time: 2-4 hours can get a first version out. Technical barrier: no coding needed; typing, uploading images, and reviewing generated results is enough. First step: open “New chat” or “Create project” in either tool. I wouldn’t rush into generating video right away. I’d write a 6-shot storyboard first.

This toolset isn’t something everyone needs. If I were still validating a product and didn’t even have a stable content theme yet, I’d be fine skipping it for now. Getting the script and customer feedback loop working smoothly would matter more first.

How I’d use it at different stages

If I were just starting out, I’d use it for a 15-second teaser and not chase a polished full piece yet. I’d only care whether anyone stops to watch. If I already had 1-2 clients, I’d turn client cases and user testimonials into short videos—using Claude to straighten out the story first, then letting GPT generate the first visual pass. If I were scaling up, I’d turn “script → storyboard → generation → revisions” into a fixed template so anyone on the team could pick it up, instead of relying on one person to be inspired at the right moment.