-
grew 2 ai characters to 30k followers each, here’s the exact stack
Posted by on February 5, 2026 at 1:07 pmsharing this because i wasted months on the wrong tools before figuring this out.
i have an ai monk actor and an ai real estate agent both sitting at around 30k followers now on tiktok and instagram.
started with sora and veo 3 and they were great at first but the problem is they have already become saturated and people can recognise them in milliseconds that they are ai generated and skip. its like everyones eyes got trained on that specific ai look.
then early this year i found cliptalk pro which i saw most of other ai talking head creators were using. its pretty much the solve for all the weird problems with ai avatars.
consistent output and can make up to 4 minutes of talking in one shot.
my workflow is flux-2-pro for the initial character image then elevenlabs voice changer feature to make my audio, i cut silences from the audio and pass it to the cliptalk talking head model. they have elevenlabs built in and let you clone your voice too but i prefer doing the voice changer route myself gives more control.
the biggest tip is use your own audio and image as input dont rely on their defaults
-
1 Reply
-
This tracks. People clock the default AI look instantly now. Using your own image + audio is the real edge. Defaults are where saturation lives. Only thing I’d add is rotate micro visual quirks every few weeks so the character doesn’t get pattern-burned. Solid stack though.
Log in to reply.