Carima: an AI influencer at 55,000 followers, and three months where nothing happened

This is a real case, with numbers you can check: the profile is @carima.of. On 18 September 2026 it had 47,010 followers. On 12 October it had 55,014. Carima is a character generated with artificial intelligence, and we say so up front because that's the point of the article.

The interesting part isn't the number. It's that for the first three months the number didn't move.

The timeline, without polish

March 2026: the account is created. The character exists, the photos exist, posts go up.

March to June: almost nothing happens. The profile grows as slowly as any new account that posts whenever it gets around to it.

June to July: the way the content is produced changes. It's the only real change in the whole period.

July to August: +20,000 followers in two months.

18 September: 47,010 followers, 230 posts.

12 October: 55,014 followers, 249 posts. Another 8,000 in less than a month.

What did NOT change

Worth saying, because it's what rules out the comfortable explanations.

The character didn't change: same Carima as in March, same face, same personality. The type of content didn't change. There was no money spent on ads, no collaboration with a big account, no single viral post dragging everything else along.

If you're looking for the trick, there isn't one.

What did change

From March to June production was manual. Every shot meant a round of generations done by hand, and most of them went in the bin: the face wasn't quite her, the pose came out wrong, the hands were off. Getting one publishable photo meant throwing away several, and every post became half a day of work.

The result of working that way isn't "bad content". It's rare content. When a post costs half a day, you don't post every day: you post when you have time, in bursts. And Instagram, which rewards frequency and regularity, does nothing with bursts.

From June and July production moved to fanlygen, a workflow where the character's identity is fixed once and from then on every shot already comes out as her. The waste collapsed. The time to get a publishable photo went from hours to minutes.

So posting frequency became sustainable. That's the real change. Not a tool that makes followers: a tool that removes the reason you stopped posting.

The lesson, which holds without us

If you're growing a virtual character and it isn't growing, the right question isn't "how do I make prettier content". It's:

how much time does my next post cost me?

Because that number decides everything else. If the answer is "half a day", you won't be consistent, and without consistency it doesn't matter how good the photos are. If the answer is "ten minutes", consistency comes on its own, and consistency is the only variable we've seen actually move the numbers.

That's true with any tool. If your current workflow makes you throw away nine generations out of ten, that's the problem to solve, not the aesthetics. (The most common reason for throwing them away is a face that keeps changing; we wrote about how to keep an AI face consistent.)

The honest caveats

It's one case. One profile growing is not a law of nature, and we're not promising it will happen to you. The only thing this case shows is that the bottleneck was production, not the character.

The first three months still count. It's tempting to read March to June as wasted time. It wasn't: in those months the character got a face, a way of talking, a first audience. The July growth landed on something that already existed.

Nobody put money into ads. If we had, we'd say so, because otherwise the comparison would mean nothing.

---

Carima also runs on Fanvue, where the same tool publishes, answers fans in chat and can offer paid content. If that's your next step, read how to run an AI model on Fanvue.