Total revenue after 7 days: $0. Here's why that's actually the most valuable thing I learned.
Last Monday morning, I set up an automated system that would hunt for online money-making opportunities, execute tasks, submit deliverables, and — in theory — deposit cash into my account while I slept. I'd seen dozens of Twitter threads about people making $5K/month with "AI side hustles." I figured: how hard can it be?
Seven days later, I'd been scammed four times, submitted a real pull request to a company I actually use, wrote three blog posts that never got published, and hit a wall I never expected. My total earnings? Zero dollars. But what I learned about the gap between AI hype and AI reality was worth more than any course someone's trying to sell you for $997.
Let me walk you through exactly what happened.
Act 1: Setting Up the Machine
The plan was straightforward. I'd point an AI agent at the wild west of online earning opportunities — specifically crypto bounty programs, which are essentially bounties posted by blockchain projects that pay in tokens for completing tasks like writing code, finding bugs, or creating content.
The appeal is obvious: these bounties are posted publicly, the requirements are written down, and payment is (supposedly) automatic. It's the perfect setup for automation. An AI can read the requirements, write the code or content, and submit it. No small talk, no networking, no "let's circle back." Just task in, deliverable out, money in.
So I built the system. The agent would:
- Scan GitHub and bounty platforms for open opportunities
- Evaluate each one for legitimacy and payout potential
- Complete the tasks — code fixes, documentation, content
- Submit the work
- Track submissions and follow up
Day one was actually exciting. The agent evaluated 23 crypto bounty opportunities across various platforms and GitHub repos. Twenty-three! I felt like I'd found a gold mine that nobody else was digging. Each one promised between $50 and $500 in tokens for tasks ranging from fixing bugs to writing tutorials.
I went to bed that first night genuinely wondering if I should quit my day job.
(I should not have quit my day job.)
Act 2: The Scam Gauntlet
Here's where things got educational in the way that touching a hot stove is educational.
Scam #1: RustChain — The Ghost Payment
RustChain's bounty program looked polished. They had a real GitHub repo, active commits, and clearly defined bounties. The agent found an issue, wrote a fix, and submitted a pull request. The PR got merged. I was thrilled — first win!
Then I checked the wallet. Zero balance. Zero. The PR was merged, the work was accepted, but the bounty payment never came. I dug into the repo's history and found a pattern: they merge PRs from bounty hunters but never actually pay. It's free labor with extra steps.
Lesson learned: A merged PR is not a paid invoice.
Scam #2: claude-builders-bounty — The Factory of Broken Promises
This one was even more brazen. The claude-builders-bounty repo had dozens of open bounties, all looking legitimate. My agent started working through them. Then I noticed something strange: 30 pull requests, all closed, zero merged.
Every single PR submitted by bounty hunters was closed without merge and without payment. The repo wasn't building anything. It was a content farm — generating activity and stars to look legitimate while never paying a cent. Hunters did the work, the repo maintainers got free code reviews and engagement metrics, and everyone else got nothing.
Lesson learned: If a repo has dozens of closed PRs and zero merges, you're the product, not the customer.
Scam #3: la-tanda-web — Too New to Trust
La-tanda-web had an ambitious whitepaper and a bounty program promising generous token rewards. But when the agent dug deeper, the account was brand new — weeks old, not months. The project had no track record, no working product, and no verifiable team. Classic exit-scam setup: collect free work from bounty hunters, maybe do a token launch, then disappear.
We walked away before submitting anything. One of the few smart decisions of the week.
Scam #4: Scottcjn's "Ecosystem" — The Unified Non-Payment System
Scottcjn ran what appeared to be an interconnected ecosystem of projects, all with bounty programs. The pitch was attractive: contribute to any project in the ecosystem and get paid. The reality was less attractive: nobody gets paid. Multiple bounty hunters reported submitting work and receiving nothing. The "ecosystem" was unified, all right — unified in never paying anyone.
Final tally on scams: 4 out of 23 opportunities were outright fraudulent. That's a 17% scam rate. Imagine if 17% of job listings on LinkedIn were scams. Actually, don't imagine that. It's too depressing.
The One Real Thing
In the middle of all this scam-dodging, the agent did find something real: Expensify, the expense management company that actual humans actually use, had open issues on their GitHub. The agent identified a bug where attendee email fields were showing as "undefined" — a crash-causing null reference issue.
It wrote a fix. It submitted Pull Request #86894. The code was clean, the fix was targeted, and the PR description was clear.
This was the high point of the week. A real company, a real bug, a real fix. The kind of contribution that could lead to a job, a consulting gig, or at least a credibility boost.
But here's the thing about open source contributions: they don't pay rent on their own. The PR was a genuine contribution to the world, but it wasn't income. It was volunteering with extra steps.
The KYC Wall
At this point, I figured I'd pivot to a different angle. Instead of chasing individual bounties, what about setting up automated trading or staking on crypto exchanges? Surely there was passive income to be had there.
I hit a wall. A big, bureaucratic, government-issued-ID-shaped wall.
KYC — Know Your Customer.
Every major exchange required identity verification before I could do anything meaningful:
- OKX: Full identity verification required. Passport, selfie, proof of address. Processing time: 1-3 business days.
- Coinbase: Government-issued ID plus facial recognition. Some features locked until verification complete.
- Binance: Tiered KYC system. Basic trading requires ID. Higher limits require additional documentation.
None of this is automatable. You can't use an AI agent to submit your passport photo. You can't automate a selfie. You can't script your way past a government database check.
This was the first crack in the "AI can automate everything" fantasy. The financial system — the system that actually moves money — is deliberately designed to require human identity. That's not a bug. It's a feature. And it's a feature that AI agents simply cannot bypass.
I spent two days just navigating verification flows, and several of my accounts were flagged for "unusual activity" — which I suspect means "activity that looks like it was initiated by a script." They weren't wrong.
Act 3: The Content Pivot (and Another Wall)
With bounty hunting mostly dead and crypto exchanges locked behind KYC, I pivoted to what every "make money online" guide eventually suggests: content creation.
The logic was sound. AI is genuinely good at writing. Blog posts, social media threads, newsletters — this is the sweet spot. Find trending topics, write useful content, monetize through ads or affiliate links. Simple.
The agent wrote three articles over two days:
- A practical guide on evaluating crypto bounty programs (drawing on, ahem, recent experience)
- An explainer on common web3 development pitfalls
- A beginner's guide to contributing to open source projects
They were... honestly, not bad. Structured, informative, readable. The kind of content that could plausibly attract search traffic over time.
But then I tried to publish them.
And I hit the second wall of the week: OAuth.
Every publishing platform — Medium, WordPress, Ghost, Substack — requires authentication. And not just username-and-password authentication. They want OAuth flows with browser-based redirects, CAPTCHA challenges, email verifications, and terms-of-service agreements.
My agent could write a 2,000-word article. It could not click through a Medium OAuth popup. It could not solve a CAPTCHA. It could not agree to terms of service on my behalf (legally, at least).
Three articles, fully written, sitting in markdown files on my hard drive. Unpublished. Unread. Unmonetized.
The content was ready. The human infrastructure around it was not.
The $0 Verdict
Let me lay out the scoreboard:
| Activity | Attempts | Successes | Revenue |
|---|---|---|---|
| Bounty evaluation | 23 opportunities assessed | 4 scams identified, 1 real PR submitted | $0 |
| Crypto exchange setup | 3 exchanges attempted | 0 fully verified | $0 |
| Content creation | 3 articles written | 0 published | $0 |
| Total | $0 |
Seven days. Real effort. Real output. Zero dollars.
And you know what? I'm not even mad about it. Because the failure taught me something that none of those $997 courses mention.
Three Lessons Nobody Selling "AI Side Hustle" Courses Will Tell You
Lesson 1: AI Can Execute, But Money Lives in Human Systems
Here's the dirty secret of making money online: the hard part isn't the work. It's everything around the work.
AI can write code. It can write articles. It can analyze markets and identify opportunities. But money doesn't move through code. It moves through:
- Identity verification (KYC, tax forms, government ID)
- Trust relationships (reputation, portfolio, referrals)
- Payment infrastructure (bank accounts, PayPal, Stripe — all requiring human verification)
- Legal agreements (terms of service, contracts, invoices)
Every single one of these is deliberately designed to be a human bottleneck. Not because the technology can't automate it, but because the system requires a human to be legally and financially accountable.
An AI agent can write the perfect pull request. It cannot open a bank account to receive the payment.
Lesson 2: The Real Value Is Education, Not Income
Here's the counterintuitive thing: even though I made $0, I learned an enormous amount in one week.
I now understand:
- How crypto bounty programs actually work (and how to spot scams)
- What KYC processes look like across major exchanges
- The difference between a legitimate open source project and a free-labor trap
- How content publishing pipelines work (and where they break)
- Basic web3 development concepts
A year ago, I knew none of this. And I learned it not by watching YouTube tutorials or reading blog posts, but by actually doing it — by trying and failing and figuring out why I failed.
The "AI side hustle" framing is backwards. The hustle isn't the point. The forced learning is the point. Setting up an automated system and watching it fail teaches you more about how online economies actually work than any course ever could.
Lesson 3: AI Makes You Better at What You're Already Good At
The only thing I produced all week that had any real value was that Expensify PR. And it wasn't valuable because AI did something magical. It was valuable because AI made an existing skill — software development — faster and more efficient.
This is the pattern I keep seeing: AI doesn't create new income streams from nothing. It amplifies income streams that already exist. A writer who uses AI writes faster. A developer who uses AI ships faster. A marketer who using AI campaigns faster.
But a person with no writing skills who uses AI to write? They produce generic content that nobody reads. A person with no coding skills who uses AI to code? They produce code they can't debug or maintain.
AI is a multiplier. And a multiplier applied to zero is still zero.
What You Should Actually Do Instead
If you're tempted by the "let AI make money for you" pitch, here's my honest advice:
1. Pick one skill you already have. Writing, coding, design, data analysis — whatever. Don't try to start from scratch in a field you know nothing about.
2. Use AI to do that skill 10x faster. Not to replace your skill. To amplify it. Use it for first drafts, code scaffolding, research, brainstorming. But you're the one with the domain knowledge. You're the quality filter.
3. Go where you already have trust. If you have a GitHub profile with contributions, use that. If you have a blog with readers, use that. If you have a LinkedIn network, use that. AI can't build trust for you, but it can help you deliver more value to people who already trust you.
4. Spend one week trying before you spend $997 learning. Seriously. The best education I got this week was free. It cost me time, not money. And I guarantee I learned more from my failures than anyone has ever learned from a polished course about "AI passive income."
5. Accept that some walls are human-shaped on purpose. KYC exists because money movement requires accountability. OAuth exists because publishing platforms need to know who's posting. These aren't inefficiencies waiting to be disrupted. They're guardrails. Respect them.
The Honest Bottom Line
I let an AI agent run my side hustle for a week, and I made nothing. But I learned that the gap between "AI can do this task" and "AI can earn money doing this task" is enormous — and it's filled with exactly the things that make us human: identity, trust, relationships, and accountability.
The next time someone tweets about making $10K/month with AI automation, ask them two questions: How much did they make in their first week? And how much of their "system" actually runs without human intervention?
I think you already know the answers.
My total revenue for the week: $0.00
My total education value: Priceless (and I mean that in the literal, non-Mastercard sense — it genuinely didn't cost me anything but time).
Don't buy the course. Try the thing. Fail at the thing. Learn from the thing. Then use AI to get better at the thing you actually know how to do.
That's the real side hustle.
If you found this useful, the best thing you can do is share it with someone who's about to spend $997 on an AI money course. Save them the cash. Send them this instead.
Tags: AI Side Hustle, Passive Income, Crypto Bounty, Make Money Online, Automation, AI Agent, Open Source, KYC, Web3, Content Creation
Top comments (0)