I realized I'm not building an AI Shorts Generator anymore
For the last few weeks I thought I was building an AI Shorts Generator.
The more I worked on it, the more I realized that's not the real value.
Anyone can generate one AI video today.
What creators actually struggle with is figuring out what will work.
So instead of focusing on generating a single video, I started building tools to generate multiple content variations from one idea:
Different hooks
Different angles
Different storytelling styles
Different visual styles
Random voices
Different subtitle styles
The goal isn't to tell creators which video will go viral.
The goal is to make it easy to create multiple versions, publish them, and let the audience decide.
I'm curious:
If you create YouTube Shorts or TikTok videos, do you usually spend more time creating content, or trying to figure out which idea is worth publishing?
I'd love to hear how other creators approach this.
I think that's an interesting shift. Generating a video is becoming easier every month, but deciding what to publish is still the hard part.
If I were using a tool like this, I'd probably value the ability to test different hooks and opening seconds more than generating entirely different videos. Those first few seconds usually have the biggest impact on whether people keep watching.
It'll be interesting to see which variations consistently perform better over time.
That's exactly the direction I'm exploring.
Initially I was focused on generating completely different videos, but the more I think about it, the more I believe the first few seconds matter the most.
I'm experimenting with templates that generate different opening hooks while keeping the core idea the same. That way creators can quickly test which introduction gets the best retention before investing time into producing more content.
I'm curious whether people would actually use this workflow regularly.
That makes sense. I don't think most creators struggle with generating content anymore—they struggle with choosing between ideas. If testing different openings becomes part of the normal publishing process instead of extra work, I can see people coming back to it regularly.
do not forget to add, different camera angles/distortions, different lenses effects,
Great suggestion.
I'm already experimenting with random visual styles, voices and subtitle variations in batch mode. Camera angles, focal lengths and lens styles could be another layer of variation without changing the story itself.
I like the idea of making every batch feel visually different by default.
The shift you described feels important because you're moving from generating content to generating experiments.
Creators rarely know which version will resonate beforehand. If your product helps them test better hypotheses instead of producing more videos, that's a much more defensible problem to solve.
I really like the way you described it: "generating experiments."
That captures what I'm trying to build much better than "AI video generator."
Instead of asking AI to make one "perfect" video, I'd rather help creators generate multiple hypotheses and let real viewers decide which one works.
I'm still figuring out the best workflow, but I think that's a more realistic approach than trying to predict virality upfront.
I'm glad it resonated.
Reading your reply left me with one question about the workflow you're designing. I'd rather explain why I'm asking in the context of your product than reduce it to a few comments.
If you're interested, what's the best email to reach you on?
Sure! My email is shortsbulk.founder at gmail dot com.
Looking forward to hearing your thoughts about the workflow.
Thanks! I’ve just sent it over.
Looking forward to hearing your thoughts whenever you have a chance.