AI Side Hustles Half-true — works only if you do the unspoken work

By the Income Reality Check Editorial Team · Updated 2026-09-13 · 7 min read

Shane Hummus ranks 10 Claude AI side hustles — and quietly admits they all need the same thing

Verdict: Half-true — works only if you do the unspoken work. The methods are real and a few people genuinely clear full-time income, but every one of them runs on traffic the video treats as a detail and then sells you separately.

Shane Hummus’s video “Top 10 Claude AI Side Hustles That Can Pay A Full-Time Income (Ranked)” (about 58,000 views when we looked) counts down from AI slop channels to what he calls “magnet money”: building a YouTube personal brand. Along the way he name-checks people making $4,000 a month selling AI prompts, $800,000 a year in corporate training, and one channel he says hit $180,000 a month. Is any of it real? Mostly, yes — and that’s exactly why the framing matters.

What the video actually claims

The pitch is that Claude removes the production bottleneck for ten different income methods, so a beginner can reach a full-time income. The ranking runs from worst to best: AI slop content (#10), building an app (#9), selling digital products like Kindle books and Notion templates (#8), corporate AI training (#7), grant writing (#6), selling a niche course (#5), AI consulting for businesses (#4), embedding Claude inside an existing business (#3), a hybrid coaching-plus-service offer (#2), and building a YouTube personal brand (#1).

Hummus is careful, to his credit, not to promise passive riches. He calls app-building “two full-time jobs,” says a full AI agency “is not a side hustle,” and warns that cold-DMing strangers is “dead.” He backs each tier with a named person: Mark Lou and “$562,000” from 23 tiny websites, Thomas Frank’s “$2.1 million” in Notion templates, Jeff Su’s claimed “$800,000 in a single year” from productivity training, community member “Josh” whose channel allegedly climbed to $180,000 a month.

Then comes the thesis. Every hustle on the list, he says, needs “the exact same thing, which is traffic.” Solve traffic once — in “under four hours a week” — and the rest gets easy. The recommended way to solve it? Attend his live training and, for “0.1% of you,” book a call for one-on-one mentorship. That funnel is the real product.

What the method actually requires

Take the digital-products tier, ranked #8 and pitched as “sleep money.” Kindle Direct Publishing is real and the royalties are exactly as advertised: KDP pays 70% on eBooks priced roughly $2.99–$12.99 and 35% below that, per Amazon’s own royalty page. What the video skips is the distribution curve. Independent surveys of self-published authors consistently show a brutal power law — a large majority earn under $1,000 a year, and the median monthly income sits in the low double digits, while a thin top tier earns almost everything. Shawn making millions is real; it’s also the far right tail, built over years and a big backlist, not a weekend and a Claude prompt.

The #10 tier — mass AI content — is where the video and reality collide hardest. Hummus says “please don’t,” and the platform agrees. YouTube’s Partner Program requires 1,000 subscribers plus 4,000 valid watch hours (or 10 million Shorts views) before you earn a cent, and in 2026 its updated inauthentic-content policy put mass-produced, repetitive AI uploads on a strike path toward removal from monetization entirely. You can be banned before you ever cross the threshold.

Now the tiers he ranks highest — consulting, training, coaching. Those numbers ($100–$300/hour for AI consultants, $1,500–$15,000 per corporate workshop) are plausible market rates. But notice what he says about them himself: bookings are “lumpy,” “the outreach alone is a full-time job,” and “people hire the trainer that they’ve already seen teach.” That last clause is the whole game. These aren’t Claude businesses. They’re audience businesses with Claude bolted on.

Here’s the piece to sit with: the income methods are legitimate, but the U.S. regulator has spent two years suing people who sold them the way this genre often gets sold. In September 2024 the FTC launched Operation AI Comply, charging outfits like Automators AI, Ecommerce Empire Builders and FBA Machine for promising “passive” AI income that “virtually none” of their clients earned. Separately, the FTC returned $2.4 million to buyers of the Lurn coaching program over deceptive money-making claims. Nothing here says Hummus is under investigation — he isn’t, as far as the public record shows, and he avoids guarantees. But the pattern the FTC keeps punishing is precisely this one: real-sounding method, single success story, upsell to coaching. (U.S. readers, this is FTC jurisdiction; U.K. readers would look to the ASA and CMA, and the logic travels.)

Who actually wins this game?

Look at the winners the video celebrates and a pattern jumps out. Thomas Frank sold $2.1 million in templates — from a second YouTube channel feeding demand from his first. “Josh” got to $180,000 a month after years of genuinely useful cybersecurity videos built his authority. Davis charges premium prices because he already spent thousands of hours breaking into management consulting. Steven’s CAD course sold because he’s a real expert in software almost nobody else can teach.

Every single one of them had an audience, a rare skill, or both before Claude entered the picture. That’s not a knock — it’s the answer. The people who win these hustles are domain experts who can generate their own traffic. Claude is a force multiplier on expertise and distribution you already possess. It is not a substitute for either.

What you’d realistically earn

Set the video’s headline figures against the base rates. If the honest KDP median is in the tens of dollars a month for most publishers, a realistic beginner outcome for digital products is something like $0–$200/month for the first six months, with maybe $500–$2,000/month achievable after a year of consistent output and marketing — and only if you’re in the minority who stick with it. Service tiers can pay faster and higher because you’re billing hours, but your first paid workshop or client is gated by whether anyone knows you exist.

Which loops back to the video’s own #1: you have to build the audience first. Hummus is right about that. He just presents “solve traffic in under four hours a week” as a settled fact when, for most people, it’s the multi-year part that decides everything. The $10,000-a-month stories are outcomes of that work, not shortcuts around it.

Who this is (and isn’t) for

This makes sense for someone with a genuine, specific skill — a certification, a career transition, a niche most people find hard — who has 8–10 hours a week and the patience to build a small audience before expecting income. If you already teach something at work, the course and coaching tiers are the most grounded on the list. It does not make sense if you’re hoping AI replaces the audience-building, or if you need money this month and are tempted by the “crank out 100 videos” tier — that path now risks a ban before it pays.

What to remember

The ranking is more honest than most of the genre, because it admits the punchline: traffic is the constraint, and none of these hustles are passive in year one. The methods work. The people cited are real. What the video compresses into a footnote — building the audience — is the actual job, and the free training exists to sell you help with it. Treat the list as a menu of legitimate businesses, then judge any paid program by whether it teaches distribution or just promises it.

If you want more on how these AI pitches tend to be structured, see our looks at 8 Claude AI side hustles that “replace” a full-time job and 5 “proven” ways to make money with AI.