Room68 · Faceless AI Research & Case Files

What faceless AI pages actually earn
OFM. UGC. AI video. YouTube automation, etc.
Everyone guesses - I track & document.
Free: the case files ↓

[Room68 · Faceless AI Research & Case Files @room68ai]

A faceless X account pulling a reported $6,300 a month has 3,800 followers, has never gone viral once, and sells a Google Sheets template.

It belongs to a laid-off engineer in Houston who turned down a $138,000 job offer in a car wash parking lot.

Eight months earlier he'd lost a $142K engineering job. Sent over 200 applications in eleven weeks. Nothing tracked them properly, so he built one — a job application tracker. Three tabs.

The stripped-down version went on Gumroad free. The full one had eleven tabs: automated interview scoring, follow-up reminders, a pipeline view like a sales CRM for job hunting. $34, raised twice since.

Then he sold it from a faceless X account. No name, no face. The bio just says what the template does and who it's for.

The entire operation:
• 3 posts a day, scheduled Sunday mornings
• 1 CTA — comment TRACKER, auto-DM fires the free version
• 3,800 followers, 2,000–4,000 views a post
• 30–60 comments on every CTA
• Nothing viral. Not once.

His audience is job seekers who are online fourteen hours a day and desperate.

Then came the rest of the ladder, all from that same original problem:
• A resume template
• An interview checklist
• A salary negotiation script

Monthly, in order: $1,100 → $4,100 → $6,300.

His line was: nobody can lay me off from a spreadsheet.

Why this belongs in a feed full of AI influencer content:

He proved the part everyone gets wrong. It was never the persona or the production. It was finding people already in pain and putting the ugliest possible fix where they were looking.

What he doesn't have is a ceiling-breaker. Three posts a day is his hard limit, because a human writes every one of them.

That's the only thing an AI persona actually adds. Not credibility, not novelty — capacity. A HeyGen or Seedance presenter, Claude drafting the variants, ElevenLabs on voice: same model, same faceless account, same product, minus the constraint that one person writes every post.

He learned that across eleven weeks of unemployment. You get it from a post.

Bookmark this. Check the link in bio — same model, with the persona, the product and the checkout already standing.

[Room68 · Faceless AI Research & Case Files @room68ai]

$9.99 a month is the entry tier on an AI OFM page, and on the operator's own accounting it is the smallest of five separate revenue lines. Nobody names the other four.

He runs a faceless AI persona — a generated character on a fan subscription platform, his own face nowhere in it — and puts his total across those five lines at over $200,000.

The subscription is the one every beginner builds for. It is also the slowest:

→ fan subscriptions — $9.99 a month, recurring, and the hardest to start
→ pay-per-view and locked sets — priced per unlock, no waiting on a rebill
→ tips and custom content — highest margin, entirely demand-led
→ brand deals — one promotional post from a top-tier virtual influencer is quoted at an estimated $10,000 to $100,000
→ ad revenue on the social accounts feeding the page — slow, and it never switches off

Put the first line beside the fourth and the shape of this business changes. It takes roughly 1,000 paying subscribers at $9.99 to match what a single brand deal can pay at the top of that range. And that fourth line is the one growing: virtual influencer spend is projected at $154 billion by 2032.

Which is why the subscription page is not the product. It is the evidence.

A brand cannot buy an audience it has no proof exists. The subscription page is where an AI persona demonstrates that strangers will pay to follow her, and that demonstration is what makes the four larger lines reachable at all. Read in that order, the $9.99 stops looking like the ceiling and starts looking like the receipt.

The entry cost matches. Character-model training ran about $24 and roughly 40 minutes, so three subscribers cover the build:

→ the AI persona and the photo sets — the job Nano Banana does now
→ the stills turned into the paid video — the job Kling does now
→ the profile copy, the captions and the fan DM drafts — Claude
→ the paywall, the subscription tiers and the pay-per-view — a settings screen

No agency. No roster. No cut of anything.

Bookmark this — the fan subscription is the smallest of the five lines and the one that unlocks the rest, which is the reverse of how it gets sold.

And if you judged this business on subscription revenue and decided the numbers were too small, that was the right maths applied to the wrong line.

In the system I use, the AI influencer, the product, the funnel, the emails and the backend all arrive already built for $27.

Check the link in bio for the whole five-line path, without an agency, a roster or a revenue split.

[Room68 · Faceless AI Research & Case Files @room68ai]

$6,000 in one month from AI OFM. The month before, the same faceless AI influencer page on Fanvue made $680. Same girl, same paywall, same 45 minutes a day. Only one thing changed.

AI OFM is OnlyFans management run with a generated creator instead of a hired one — the same fan-subscription business as OnlyFans, Fansly, LoyalFans or ManyVids, except the AI girl on the page was never a person.

What changed was how many videos went out. He made 57 short clips in one day, then posted 4 every day and waited.

The two months, side by side:

→ month one: $680, first paying subscriber on day 6
→ month two: an estimated $6,000, close to nine times more
→ videos made in one day: 57
→ videos posted per day: 4
→ paid subscribers on the fan-subscription page: 57
→ his first video that worked: 11,500 views
→ the one that changed the month: 675,000 views

Look at those last two numbers. 11,500 and 675,000. Same AI girl, same page, same afternoon. One did almost nothing. The other paid for the whole month.

No way to tell them apart before posting. Not for him, not for anyone.

Nobody picks the winner in advance. The only lever is how many tries are in the queue. He stopped guessing and posted 4 a day until one hit.

So month one was not a failure. $680 only looks like one. It was the pile being built, and the 675,000-view video could not have landed in month one, because month one did not have enough videos yet.

That is where most AI model pages die. Three or four weeks in, small numbers, and quitting feels sensible.

Making the clips is not the slow part now. Claude writes the character and the on-screen text. The job Nano Banana does now makes the photos. The job Kling does now turns a photo into a moving clip.

The paywall side is the same. An OnlyFans agency hires chatters to answer subscribers; here Claude drafts the replies and prices the pay-per-view sets.

No filming. No studio. No crew.

The slow part is waiting to find out.

Bookmark this — $680 in month one is not a verdict on the page, it is a pile that is not tall enough yet.

And if something got shut down around week three, that was not a discipline problem. The feedback was simply too slow to say anything. A page was asked to prove itself before enough videos existed for one of them to land.

In the system I use, the AI influencer, the offer, the funnel, the customer service and the closing all arrive already built for $27, so the weeks normally spent making them are spent posting instead.

Check the link in bio if the part you keep losing to is the long wait before you know whether any of it works.

[Room68 · Faceless AI Research & Case Files @room68ai]

@neilpatel Cited

[Room68 · Faceless AI Research & Case Files @room68ai]

@wannercashcow Megaprompt

[Room68 · Faceless AI Research & Case Files @room68ai]

@BrunoF566 System

[Room68 · Faceless AI Research & Case Files @room68ai]

@ecompol Pol

[Room68 · Faceless AI Research & Case Files @room68ai]

Faceless AI pages get judged on follower count.

The number that actually predicts revenue is how many pages are running, not how big one gets.

41 pages. One of them made a third of the money. Nothing marked it in advance.

[Room68 · Faceless AI Research & Case Files @room68ai]

@jasonugc System

[Room68 · Faceless AI Research & Case Files @room68ai]

@X_FINALBOSS Blueprint

[Room68 · Faceless AI Research & Case Files @room68ai]

@X_FINALBOSS How

[Room68 · Faceless AI Research & Case Files @room68ai]

The quietest flex in 2026: doing the work of a team of five, alone — and never once explaining how.

Some people are already living it. Most haven't noticed yet.

[Room68 · Faceless AI Research & Case Files @room68ai]

Old way: hire people to scale.
New way: build a system, and scale yourself.

The quietest flex of this era is doing a lot, alone, and never explaining how.

[Room68 · Faceless AI Research & Case Files @room68ai]

A single person can now quietly do the work of a five-person team. No employees, no office, no one the wiser.

Here's how the setup actually works 🧵

[Room68 · Faceless AI Research & Case Files @room68ai]

This is why a faceless page can feel like a whole media company.

Behind it: one person, a few tools, and a process that runs on rails. It looks like a team because the output is team-sized.

[Room68 · Faceless AI Research & Case Files @room68ai]

An uncomfortable question:

If a stranger with your exact resources but zero fear started today, how far ahead of you would they be in 90 days?

[Room68 · Faceless AI Research & Case Files @room68ai]

You're closer than you think.

Most people quit at the exact point where it was about to get easy.

[Room68 · Faceless AI Research & Case Files @room68ai]

Which is your real bottleneck right now — ideas, time, or just starting?

Curious what most people say.

[Room68 · Faceless AI Research & Case Files @room68ai]

You don't need more motivation. You need one boring habit and a tool that makes the boring part disappear.

Both exist. Most people still won't do it.