Loading page…
A talking-head account whose presenter does not exist: a HeyGen avatar, an ElevenLabs voice and five minutes in CapCut per video, posted relentlessly — 2.5 million followers, over two million of them gained in under 90 days, at a stated $50,000 to $70,000 a month in profit
2.5M followers, over 2M gained in under 90 days · a sample video past 300,000 views · production tooling under $20/month (creator-reported by Richard Yu about a friend, July 2026)
"You might feel it's real, but it's 100% made by AI." The presenter, the voice and the delivery are all synthetic; the only human inputs are the script idea and five minutes in CapCut on a phone.
Fewer bars = easier, cheaper, or faster for an AI-assisted solo builder. Editorial judgments based on the case details.
Midway through a free 44-minute course on selling digital products, Richard Yu pulls up an account belonging to a friend he calls Young Manasseh and reads out the numbers: 2.5 million followers, more than two million of them gained in under 90 days, and "consistently anywhere from 50,000 to 70,000 US dollars in pure profit on a monthly basis." He opens one of the videos, which has over 300,000 views. A presenter talks straight to camera about psychology — "people who overthink everything, they're not overthinking just because they're anxious, they're overthinking because their mind was trained to look for danger" — and then Yu pauses it. "You might feel it's real, but it's 100% made by AI." The presenter is a HeyGen avatar, which Yu says costs under twenty dollars a month and has a free tier. The voice is ElevenLabs. The captions and the cut are done in CapCut on a phone. "This video takes less than 5 minutes to actually make, and he posts simple videos like this, which is what grew him over 2 million followers." That is the entire disclosed operation: a content format with no filming, no studio, no face, and a per-unit cost close to zero, run at volume. What Yu does not say is how the money is made — the profit figure arrives with no product, price, funnel or platform attached to it, which is the single biggest hole in this case and is treated as such throughout this entry. The surrounding course lays out the content method Yu teaches: find five accounts in your niche, take their five best recent posts, study the hooks rather than copying the videos, post one to three times a day on a single platform, and turn whichever short video overperforms into a longer one.
For an account like this the first customers are viewers, and they are won by hooks rather than by reach.
The quick-reference card — acquisition channels, replication playbook, risks & traps, and our verdict — is members-only.
Join the Unicorn Club — $5/moCancel anytime, in one clickPaste into Claude Code / Codex and get a working version of this product end-to-end.
Two production-grade prompts per idea — a full build spec and an SEO growth plan — plus 10 tailored build specs a week. Members only.
Join the Unicorn Club — $5/moCancel anytime, in one clickEvery claim traces back to a listed source — open any of them and check the number yourself.
Data credibility: Creator-reported, secondhand. Richard Yu presents this account in a free full-length course video on his own YouTube channel (46,060 views at capture), describing Young Manasseh as a friend. The figures — 2.5 million followers, over two million gained in less than 90 days, $50,000 to $70,000 a month in pure profit, a sample video above 300,000 views — are all spoken by Yu, not by the operator, and no analytics screen, payout statement, account handle or platform name is shown on camera. The name is transcribed from auto-generated captions and may be spelled differently in reality; the account is never displayed, so none of this can be looked up. The production details (HeyGen under $20/month with a free tier, ElevenLabs for voice, CapCut on a phone, under five minutes per video) are the most reliable part of the claim because they are demonstrated elsewhere in the same video. Note the incentive: the video is monetised by affiliate links to two software products with negotiated discounts, offers a $997 course as a signup bonus, and ends by routing viewers to a paid program — so the case study functions as marketing for an education business. Yu also states his own business does over $500,000 a month, which is covered separately in this directory and is not evidence for this entry. No monetisation mechanism for this account is disclosed anywhere. ProvenStartups has verified none of it.
Attention arrives first in this model and money arrives second, which means the first dollar comes from a conversation rather than from a launch.
The compounding here comes from production cost collapsing to near zero.
The dead stretch for a faceless avatar account is the first thirty to sixty videos, when uploads land at a few hundred views and nothing suggests the next one will be different.
On the evidence in this video, the product is an attention machine with an undisclosed back end. The visible half is precise: a synthetic talking-head presenter delivering short psychology insights, captioned, vertical, posted daily, with hooks that name a feeling the viewer recognises in themselves. The sample line is a good specimen of the genre — it reframes something the viewer already experiences (overthinking) as something other than a personal failing, which is exactly the trade that makes this niche travel: the viewer feels explained rather than instructed. Our read: the niche choice is doing more work here than the AI tooling is. Psychology and self-understanding content has no seasonality, no geography, no expertise requirement the audience can check, and an effectively unlimited supply of topics, which is why it is one of the few categories where a synthetic presenter is not a disadvantage — nobody needs the speaker to be a credentialed clinician for a 30-second reframe. The invisible half is the business, and it is genuinely invisible. No product, price, platform, landing page or funnel is named anywhere in the source, so anyone copying this is copying a content format with a profit figure attached to it by a third party, and should build their own monetisation from first principles rather than assume one exists to be reverse-engineered.
This is the honest gap in the file. The transcript states $50,000 to $70,000 a month in pure profit and stops there — no product, no price point, no platform, no split between creator fund payouts, brand deals, affiliate revenue and the creator's own digital product. Our read: at 2.5 million followers in a psychology niche, all four are plausible and they have wildly different economics. Platform payouts on short form are volatile and per-view, so they scale with uploads but not with audience quality; brand deals pay well but require an operator willing to negotiate and a niche advertisers want; affiliate revenue is the easiest to start and the least durable; a self-published digital product is the only one where the margin is yours and the audience relationship is an asset. The surrounding course pushes viewers toward the last option — a one-time guide or a recurring micro-app — and that is the version worth modelling, because it is the only one where you own both the traffic and the offer. If you copy this, decide the monetisation before the first video rather than after the first million views, because the offer determines which hooks are worth chasing: an audience assembled on generic psychology hooks is much harder to sell anything specific to than one assembled on the exact problem your product solves.
The disclosed stack is three tools and a phone. HeyGen generates the on-screen presenter, and Yu puts it at under twenty dollars a month with a usable free tier. ElevenLabs generates the voice. CapCut on the phone handles the cut and burns in the captions. Production time per finished video is stated as under five minutes. That is the entire cost base for the content side, which is why the word used for the earnings is "profit" rather than revenue — at this scale the monthly tooling bill is a rounding error against a five-figure month. Our read: the numbers imply something worth stating plainly, which is that the cost structure of short-form content has changed category. A conventional faceless channel still pays for stock footage, an editor, or the creator's own filming time; this one pays for neither talent nor time, so the marginal cost of testing a new format is close to zero and the correct strategy is therefore volume and iteration rather than polish. The real costs sit elsewhere and are not in the transcript: whatever platform, storefront and payment processing the back end runs on, and the operator's own hours spent writing scripts and answering DMs. Budget for a small paid tier of each AI tool, a domain, and a checkout — call it under a hundred dollars a month all-in — and treat everything above that as evidence you are buying your way out of a content problem rather than solving it.
Copy the production discipline, which is the genuinely transferable part: a synthetic presenter that never needs scheduling, a five-minute pipeline, one platform, one to three posts a day, hooks researched from twenty-five proven posts refreshed monthly, and every overperforming short expanded into a long-form piece. Copy the niche logic too — pick a category where the audience needs no credential from the speaker and the topic supply is infinite, which is why psychology, habits and self-understanding work for synthetic presenters while medical, legal and financial advice do not. Change two things. First, build the offer before the audience: this file cannot tell you how the money was made, and an audience without a product attached is a hobby with good analytics. Decide on day one whether you are selling a low-priced guide, a recurring micro-tool or affiliate placements, then choose hooks that attract people who would buy that specific thing rather than hooks that maximise views. Second, disclose the synthetic presenter. Our read: platforms are moving steadily toward labelling AI-generated content, and an account whose entire equity is a face that does not exist is safer if it says so up front than if the disclosure is made for it later.
The evidence risk comes first and is large. Every number here — 2.5 million followers, two million in under ninety days, $50,000 to $70,000 a month in profit — comes from Richard Yu describing a friend's account inside a video whose purpose is to route viewers to affiliate links and a paid program. No dashboard, analytics screen, payout statement or account handle is shown, and the platform is never named. Treat this as a format to study and a claim to test, not as a proven income figure. Second, platform concentration: one account, one recommendation algorithm, no owned audience mentioned anywhere. A ranking change or a suspension takes the whole business at once, which is why an email list or an owned storefront should be built in parallel from the first week. Third, synthetic-media policy. Our read: rules on AI-generated presenters, labelling requirements and monetisation eligibility are being rewritten continuously, and an account with no human on camera is exactly the kind that gets caught by a policy change it did not see coming. Fourth, saturation — the tooling costs twenty dollars and the method is being taught in a free video with 46,000 views, so the format's advantage is time-limited and the differentiator collapses back to writing. Fifth, the missing back end: without a disclosed monetisation path, a copier can reproduce the audience and still earn nothing.
The four founder playbooks — first customers, first dollar, the flywheel, the silent stretch — are in the Unicorn Club, on this idea and on all 1,435 others. The story, the numbers, the sources and every later section stay open to everyone.
Join the Unicorn Club — $5/moCancel anytime, in one click