I Tested 5 AI Side Hustles. This One Made Me $35,677

 

I spent the better part of a year throwing AI at every side hustle people swore would make me rich.

Build an app. Sell websites. Sell automation to local businesses. The internet was loud about all of it, and the pitch was always the same: make easy money with AI!

Most of those ideas went basically nowhere. But one of them has made me $35,677 and counting.

In this article, I’ll share exactly what I tried, why most of it flopped, and the one thing that actually worked.

Attempt #1 - Building an App With AI

This was the dream, right? Describe an app, let AI write the code, ship it, collect subscriptions while you sleep.

I actually built the thing. The AI got me most of the way there faster than I ever could’ve done alone. The problem wasn’t the building. The problem was everything that came after building.

Nobody knew the app existed. Distribution is the entire game, and AI doesn’t hand you that. I had a working product, and nobody cared. I also hit the wall every no-code-turned-real builder hits: the last 20% of an app — edge cases, payments, bugs that only show up with real users — is where the actual work lives, and AI got noticeably worse the deeper I went.

I had built a solution and gone looking for a problem. (backwards)

Attempt #2 - Selling AI-Built Websites

Next up: spin up websites with AI, sell them to small businesses for a few hundred bucks each. Volume play. Sounded clean.
It’s a real business. It’s also a grind that nobody mentions in the YouTube thumbnails. Selling websites means cold outreach, follow-ups, scope creep, revisions, and clients who go dark for two weeks then want everything changed by Friday.
The AI made the websites. It did not make the sales calls, chase the invoices, or handle the “can you just move this button” emails. I can do sales. I don’t love sales. And this was all sales.
I tapped out before it ever became real money.

Attempt #3 - AI Automation to Local Businesses

This one had the best story attached to it. Walk into a local business, find a boring repetitive task, automate it with AI, charge a setup fee plus a monthly retainer. Recurring revenue. Beautiful on paper.
In practice, the sales cycle is brutal. Local business owners are busy, skeptical, and not exactly waiting around to hear a stranger explain large language models.
Every deal required a ton of education before anyone would even consider paying. And once you build the automation, you own it. When it breaks, your phone rings. I’d traded a job for a bunch of tiny demanding clients who each thought they were my only one.
Real opportunity here. Wrong fit for someone with 10 to 20 hours a week and no appetite for that kind of hand-holding.

Attempt #4 - AI “Faceless” Content Channels

I’ll keep this one short because it was the shortest-lived. The plan: AI-generated faceless videos, post at volume, ride the algorithm.

The output was generic. The internet is already drowning in soulless AI content, and the algorithm knows it. I was making things that looked like everyone else’s things. Nothing connected because there was nothing of me in it.

Additionally, the platforms have cracked down on this AI side hustle.

Attempt #5 - AI Digital Products

Here’s where it turned around.

Instead of using AI to build something impressive, I used it to solve one specific, annoying problem for one specific group of people. I packaged the solution into a simple digital product and put it on Gumroad.

That one product has made over $20,000.

No app store. No cold calls. No retainer clients texting me at 9pm. Someone has a problem, they find my product, they buy it, they solve their problem. That’s the whole loop. And it runs whether I’m at my desk or asleep.

Once I saw it work, I did the obvious thing: I expanded into other niches with the same playbook. Find a real problem, package the solution, sell it. To date, that approach has made me $35,677 across products.

Why This Works

The first four all had the same flaw. I was leading with the technology and hoping a problem would show up. “I built an app.” “I made websites.” “I set up automation.” Cool — for who, and what pain does it address?

The digital product flipped it. I started with a problem people were already trying to solve, and AI just helped me build and ship the solution faster.

That’s the actual unlock. The money isn’t in AI. The money is in solving problems. AI is the fastest tool we’ve ever had for building solutions — but it can’t pick the problem for you, and it can’t make people care.

A few things the digital product had that the others didn’t:

Zero ongoing obligation. Once it’s made, it sells. No clients, no support contracts, no 2am outages.

Built-in distribution. I already had an audience that trusted me. The product served them instead of chasing strangers.

Margins that don’t quit. A $20 product costs me nothing to deliver the 500th time.

I found something that laid out the ENTIRE framework, not just product creation, but the sales psychology, the platform algorithms, the pricing strategies that actually convert. The Digital Profit Blueprint was honestly the first thing that didn’t feel like I was collecting puzzle pieces from seventeen different sources. It’s like someone finally turned the lights on in a room I’d been stumbling through in the dark. Everything from setting up your store to making those first sales using strategies that aren’t just “post daily and hope,” it’s all there, step by step. The EXACT process.

Stop asking: “What can I build with AI?”

Start asking: “What problem can I solve, and how can AI help me solve it faster?”

Four of my five experiments failed because I had the question backwards.

The tools will keep getting better. They’ll keep tempting you to build flashy things nobody asked for. Ignore that. Find the problem first.