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

By the Income Reality Check Editorial Team · Updated 2026-08-07 · 7 min read

How a 30-year-old made ₹1 crore with Claude Code — and what the video skips

Verdict: Half-true — works only if you do the unspoken work. The money is real, but it rode on a decade of sales experience and a big prior audience, not on the AI tool.

On episode #251 of The Sanskar Show, guest Sandy Lee tells host Sanskar Goyal she made “over 90k USD in the last two months” using AI — the headline rounds that up to ₹1 crore (~$120,000). She’s a mother of three with a full-time job, and she credits one tool above all: Claude Code. Is that plausible? Mostly. But the version a beginner hears — “learn Claude Code, build an app, get paid” — leaves out the machinery that actually moved the money.

What the video actually claims

Sandy is candid, and that’s what makes the interview worth reading closely. She lays out her revenue as several “income baskets.” The first was a $3,000/month retainer for “SEO blog automation” that she landed on Upwork within two weeks of applying — despite, by her own account, knowing nothing about SEO going in (“fake it until I make it,” she says, after cramming with AI). Later, inbound leads turned into consulting gigs, including a $13,000 client who hired her as an AI content strategist.

From there the baskets multiply: YouTube AdSense, brand sponsorships (her first came at 9,000 subscribers), a paid Skool community, and small web apps she built “100% with Claude Code” — an SEO audit tool, a content-automation tool — one of which she prices at $2,000. She’s on Claude’s $100/month Max plan, uses Opus for planning and Sonnet for execution, and saves repeat workflows into markdown memory files and custom skills so the AI edits her videos and drafts her scripts.

Her thesis, stated plainly: everyone needs a personal brand, and Claude Code is the one tool that let a non-technical person run several businesses in the two or three hours she has after the kids are asleep.

What the method actually requires

Here’s the line most viewers will skim past. Sandy says she’s “been a content creator since 2018” and grew a channel “to 550K across three different platforms” teaching Korean to Spanish speakers, plus a separate Instagram she took to 150K. She spent 10 years in sales and marketing. She’s paid for “a lot of different programs in the past.”

That is not a beginner. That’s a seasoned marketer with a distribution engine and a decade of client-facing sales reps, using AI to move faster. The AI is real leverage — but leverage multiplies what you already have.

Look at the Upwork story. Landing a $3,000 retainer in two weeks with zero domain knowledge is not the median experience. Upwork’s own resource on freelancer pay notes that while skilled U.S. freelancers can do well, a large share sit far lower — Upwork’s data puts the typical hourly rate near $39, and independent tallies of the platform show roughly 55% of freelancers earning under $50,000 a year and about 30% making under $15/hour. For context, the U.S. Bureau of Labor Statistics pegs the national median wage around $28.56/hour. A $3,000 close in week two is the top of the distribution, not the middle of it.

The tool cost is the one number the video gets exactly right, and it’s small. Claude Code runs $20/month on Pro and $100 or $200/month on Max, per Anthropic’s pricing. Sandy pays $100. So the AI isn’t the expensive part — your time and your audience are.

Notice the pattern? Every basket assumes an audience or a sales skill is already there. The Claude Code app didn’t create demand — Sandy says outright she “found people’s need first before I build a product,” and that need surfaced because her videos were already getting views.

Who actually wins this game

The people who replicate a result like this fall into familiar buckets: they already have an audience, they have real sales ability, or they were early. Sandy has the first two in abundance. Her YouTube monetization didn’t start from nothing either — under YouTube’s Partner Program rules, you need 1,000 subscribers plus 4,000 public watch hours in a year (or 10 million Shorts views in 90 days) before ad revenue turns on at all. Her sponsorship at 9,000 subs sounds early, but 9,000 engaged subscribers is itself the product of years of practice at hooks, formats, and posting cadence — the exact skills she spends half the interview describing.

She’s also transparent about a cost most “went viral” stories hide: one Instagram post blew up, 200-plus people wanted a link, and automating those DMs through ManyChat would have run roughly $2,000 for that single post before she switched to a cheaper tool. Going viral can cost money. That’s a detail beginners rarely hear.

What you’d realistically earn

If you’re genuinely starting from zero — no audience, no sales background — the honest range for the first six months is close to nothing. Zero to maybe a couple hundred dollars, mostly from one small freelance gig if you’re good and persistent. A year of consistent output might get a focused person into the few-hundred-to-low-thousands-per-month band, and even that assumes real skill-building along the way.

Sandy’s own advice quietly confirms this. She warns against “shiny object syndrome,” tells you to pick one thing, and says the first $100 comes from consistency, not a magic tool. That’s sound. It’s also the opposite of the “AI made me $90k” framing the title sells. The $90,000 is a two-month peak from someone with a decade of runway behind it — not a monthly floor, and not net of her time, her ad-tool bills, or her sponsorship discipline.

A note for U.S. readers specifically: the FTC is tightening the rules here. In 2025 the agency proposed a new Earnings Claim Rule that would require written substantiation for income claims in money-making and coaching offers. Sandy isn’t selling a get-rich course, and nothing here suggests she’s crossed that line — but the regulator’s stance is a useful filter for the flood of “I made $X with AI” videos you’ll see next.

Who this is (and isn’t) for

This path fits someone who can commit two-plus focused hours a day for months, who has (or is willing to build) a sales or content muscle, and who treats AI as an accelerator rather than a substitute for distribution. If you already have an audience anywhere — a mailing list, a following, a warm client network — Claude Code genuinely compresses the work of turning that into products and services.

It’s a poor fit if you’re hoping the AI replaces the audience-building and the selling. It won’t. If you can’t yet name who would pay you and why, the tool has nothing to accelerate.

What to remember

Sandy Lee’s numbers look real and her tactics are specific and useful — the markdown memory, the custom skills, the “find the need before you build” discipline. The half-truth is in the causation. Claude Code multiplied a business that already had a marketer, an audience, and ten years of sales behind it. Buy the $100 tool if it helps; just don’t buy the idea that it’s the reason. Want the same result? Start with the boring part she rushed past — one skill, one audience, done consistently. For more on this pattern, see our breakdowns of Claude Code YouTube income claims and Claude AI side hustles that can pay a full-time income.