damn i think i understand how ai ugc works

I always kind of dismissed AI UGC as being a waste of time (and money) because the amount of effort to create a following on multiple AI influencer channels feels like a lot

but that's not what people are doing.

brands are getting the videos and running it as ads / trial reels / tiktoks on secondary accounts.

this actually makes sense.

instead of engaging an influencer for $1000-2000/video, you'd get an AI influencer to make explainer videos about your product and do this at scale so produce 50-100 videos in a week.

then run everything, test, change scripts, hyperpersonalise for different ad regions and segments, and optimise.

this is where platforms like weavy or higgsfield kind of make sense, where they help to create all of these. though the future would probably be something with a claude code-like experience.

you'd upload all the real product b-roll, then have the AI characters generate short talking head / reaction parts, then stitch everything together with remotion or cardboard.

ok but isn't that unethical

I think stealing an identity of a real person and making UGC videos with them is definitely not cool.

though it's a bit of a grey area in this sense. Yes you are reducing opportunities for creators to be getting these kinds of UGC things. but 1) capitalistic markets makes this kind of inevitable and 2) the value proposition changes.

what I think will happen is UGC creators are going to be valuable because of the long term brand that they build, not the transactional video that comes across.

Working with a creator with 100k followers in your niche would still be wayyy better than a fake ugc creator that nobody knows.

so as it is with most things, the brand is what becomes valuable and is what's worth building, not really the content itself. Creators need to switch from the service provider model to the trust + community model.

the future

plus, I can imagine a future where some leading UGC creators will offer an option where you can test 20 versions of their UGC video (AI versions that the creator made in 1 hour) then for the best performing one - they'll create a better video leveraging the data on what worked.

I think brands would do this too, run 100 ads with different AI UGC creators, and the ones with the highest conversion rates & lowest CPM, they just ask real creators to reference these videos and then double down on the human-made ones with recognizable creators in various niches.

and of course, a saas opportunity

platforms like weavy are cool and give you lots of control to get consistent high quality videos generated, but combining weavy + remotion / cardboard --> this would be insane.

the barrier right now is you need to take your weavy generated videos and then go to premiere pro/davinci to stitch them together and edit everything before exporting with lots of tedious export steps.

I think the future will be a new capcut that’ll emerge which combines the weavy step of leveraging all the ai models with video rendering and ai editing too, with an orchestrator agent that can manage the end to end process.

Maybe it’ll be built within premiere pro, but probably not. It’s not a trivial task for a startup though so would be cool to see if someone pulls this off.