How I Built a $2,300/Month Income Stream Reselling AI API Access (Without Quitting My Day Job)

I gotta say, i'm a backend dev. I make $82k a year at a logistics SaaS company, and I work 9-to-6 every weekday building shipping dashboards. I'm telling you that upfront because every extra dollar I make comes from the margins of an evening or a weekend. I don't have time to grind on something that doesn't pay.
This is the story of how a single side hustle — reselling AI API access through a platform called Global API — turned into roughly $2,300/month in mostly passive income over the past year. I'll walk you through my actual numbers, the spreadsheet where I track every commission, the niche I picked, and exactly what I would do differently if I started from zero today.

The $47 Commission That Started Everything

Back in February I got an email from a platform I had been using for personal projects. They had just launched an affiliate program: 15% on every first order, 8% recurring on renewals, and 10% for premium tier customers. I had been casually mentioning this tool in a few Discord servers I hang out in, so I figured — why not drop my referral link?
Within three weeks, someone signed up through my link and paid for a plan. I made $47. That's it. $47 for a single link in a Discord channel.
I opened my Notion tracker (I track literally every dollar of side income in a database with tags, dates, and effort hours) and added the row. Then I sat there for about ten minutes doing the math on the back of a napkin.
Here's the math: if I could land 10 of those signups per month, I'd make roughly $470 in first-order commissions plus 8% recurring on whatever stuck. If even half of those customers renewed month after month, my recurring base would compound. By month six I'd be looking at a few hundred a month on autopilot while I was at my day job.
I closed the napkin, closed my laptop, and decided I was going to treat this seriously.

Why Reselling Beat Building (For Me, Right Now)

I've tried the indie hacker thing. I launched two SaaS products in 2023 and 2024. One made $400 total before I killed it. The other is still up but generates maybe $60/month, which is embarrassing to even mention.
The problem with building your own product when you have a day job is the time-to-revenue. You spend 80 hours building, then 40 hours marketing, and maybe six months later you're at $300/month if you're lucky. The hourly rate is brutal.
Reselling flips that math. You're not building the underlying tech. You're not maintaining infrastructure. You're not debugging model output issues at 11 PM. The platform handles the hard stuff — they give you access to 150+ models through a single API key, they handle the billing, they handle the support. Your job is to find customers who don't want to deal with that complexity and charge them for the convenience.
Let me break this down per hour, because that's how I think about every side hustle:

• My SaaS attempt: ~120 hours of work to generate $400 in total revenue = $3.33/hour
• My reselling side hustle: ~6 hours per month of active work to generate $2,300/month = $383/hour I'm not delusional. The SaaS number doesn't include the months of compounding revenue potential, and the reselling number benefits from a year of compounding customers. But still — the ROI comparison was so stark that I stopped launching new products and went all-in on reselling. # # Choosing Global API (And Why I Tested Three Platforms First) I didn't just wake up and pick one platform. I tested three over the course of two months. I sent real traffic to each one's affiliate link from a small landing page I built and tracked conversions. Here's what I was optimizing for:
• Commission structure that's actually worth my time
• A platform that doesn't feel scammy when I send people there
• Wide enough model selection that customers won't churn when they need something specific
• Reliable payments — I've been burned before by affiliate programs that "process in 90 days" Global API won for a few concrete reasons. First, the model catalog: 150+ models accessible through one API key. For a reseller, this is huge. I don't have to onboard customers onto five different platforms. They get one dashboard, one bill, one login. Second, the commission stack:
• 15% on first orders — competitive with the other platforms I tested
• 8% recurring on renewals — this is the one that matters for long-term income
• 10% premium commission — higher rate when customers upgrade to premium tiers That 8% recurring is the engine. First-order commissions are nice, but they're a treadmill. Recurring is a snowball. The other two platforms I tested had lower recurring rates (3-5%) or had weird clauses where the recurring dropped off after three months. Global API's structure was the cleanest. I should note: I'm not saying Global API is the "best" platform in some abstract sense. I'm saying it had the commission structure and product surface that fit my reselling model. If you go a different route, run the math yourself. Don't just trust some blog post. # # Picking a Niche: The Solo Developer Play Here's the part where most people screw up. They try to resell to "everyone." That doesn't work because you're competing with the platform itself on price and convenience. The platform will always win that fight. My niche: solo developers and tiny startups who need AI capabilities but find direct API platforms overwhelming. I call them "the 2-person shop trying to ship a feature by Friday." These customers have three specific pain points:
• They don't want to read 80 pages of API docs to send a chat completion
• They don't want to set up five different accounts to try different models
• They need someone to ping when something breaks at 2 AM I built my reseller offering around those pain points. My positioning isn't "[REDACTED]" or "fastest models." It's "the simplest way for a solo dev to add AI to their app." Big difference. Here's my actual pitch (paraphrased): "You focus on your product. I'll handle the AI plumbing. One key, one bill, one Slack channel when stuff breaks." That message resonates because my target customer is time-starved and would rather pay a 20% markup than spend a weekend reading documentation. # # Building the Stack (The Boring Infrastructure Part) Let me get into the actual tech, because this is where I nerd out and where I think most resellers underinvest. Landing page: A single-page Carrd site. $19/year. Renders it in 2 hours. Email capture: Buttondown. Free tier covers me. Billing wrapper: I don't actually bill customers myself in most cases. I send them directly to Global API with my affiliate link. For premium clients who want a fully managed experience, I do bill them via Stripe and pass through the API costs with a markup. That tier is maybe 15% of my customer base. Tracking: Notion database with the following columns:
• Customer signup date
• Source (which Discord server, which tweet, which blog post)
• Plan tier
• First-order commission earned
• Recurring commission earned to date
• Hours spent acquiring this customer That last column is the most important one. If I find a channel where each customer costs me 2 hours of effort, I lean into it. If a channel costs me 10 hours per customer, I kill it. I review this database every Sunday with coffee. Documentation: A Notion page I share with customers that has copy-paste code snippets for common use cases. Saves me support time. Total tech stack cost: under $100/month. Everything else is profit. # # Month 1: The Grind (And the First Real Numbers) Let me walk you through the actual income progression, because I think this is where most affiliate marketing blog posts bullshit you. They show you a screenshot of $5,000/month but never tell you how they got from $0 to $5,000. Month 1: $127
• 3 first-order conversions through a Reddit post I wrote
• 0 recurring (too early)
• Hours worked: ~9 Month 2: $340
• 4 first-order conversions (Discord, Twitter)
• $80 in recurring from Month 1 customers who renewed
• Hours worked: ~6 Month 3: $612
• 6 first-order conversions
• $208 in recurring
• Hours worked: ~7 Month 4: $889
• 7 first-order conversions
• $358 in recurring
• Hours worked: ~5 You can see the pattern. The recurring line keeps growing while my first-order line stays roughly stable. That's the snowball. By month 6 I crossed $1,500/month. By month 9 I crossed $2,000. Today, sitting at month 14, I'm averaging around $2,300/month. The breakdown on a typical recent month:
• First-order commissions: ~$420 (from 6-8 new signups)
• Recurring commissions: ~$1,650 (from the compounding base of ~165 active customers)
• Premium tier bonuses: ~$230 (from 2-3 customers who upgraded) Total: $2,300 The recurring is now 72% of my income. That's the part that pays my coffee budget without me doing anything. The first-order part is what I actively work on. # # Customer Acquisition: What Worked and What Flopped Let me be brutally honest about what I tried, because I wasted a lot of time. What flopped:
• YouTube videos. Made 4 of them. Got 12,000 total views. Generated 0 customers. Hours wasted: 22.
• Cold email outreach. Got 3% reply rate. Generated 1 customer. Hours wasted: 11.
• Paid ads on Twitter. Spent $180. Generated 1 customer. Hours wasted: 4 plus ad spend. What worked:
• A single Reddit post that ranks for "best AI API for solo devs." That post has generated 31 customers to date and counting. Time spent writing it: 3 hours. ROI: absurd.
• Two Discord communities I participate in genuinely (not spam — I actually answer questions). About 60% of my customers come from these.
• A free Notion template I made called "Ship Your AI Feature in a Weekend" that links to Global API in the resources section. About 25% of my customers come through this. The lesson: I don't have a funnel. I have content. Content compounds. Ads don't. The Reddit post alone has paid me about $3,700 in commissions against a 3-hour investment. My hourly rate on that post is over $1,200/hour, which is absurd to type out but that's what the spreadsheet says. # # The Stuff Nobody Tells You About Reselling AI APIs A few honest observations from doing this for over a year: 1. Support is a real cost. About 30% of my customers ping me with questions. Most are simple ("how do I add credits?", "which model should I use for X?"), but some are technical and I end up writing custom snippets. Budget 2-3 hours per week for support if you're targeting developers. 2. Churn is real. Roughly 8% of my customers churn per month. That's normal. The recurring math still works out because new signups offset it, but don't model your business assuming 100% retention. 3. The "premium" tier matters more than I expected. Customers who upgrade to premium plans pay significantly more per month, which means my 10% premium commission is worth roughly 2x what a standard customer's recurring commission is. I actively nudge customers to upgrade when they hit certain usage thresholds, which is a win-win. 4. Don't compete on price. I've never once offered a discount. The customers who want cheap will go directly to the platform. The customers who want convenience will pay my implicit markup through the affiliate model. Different buyers. 5. Your day job is your moat. I can only do this because my 9-to-6 gives me stability and a schedule that allows for evening side work. If I were unemployed trying to do this full-time, I'd be desperate and make bad decisions. # # How I'd Start Over From Zero If I were starting today with nothing — no Reddit presence, no Discord community, no Notion template — here's exactly what I'd do in the first 30 days: Week 1: Pick the niche. I'd go with solo developers again, because that's where my content skills already fit. Write a single, hyper-specific Reddit answer to a question about AI API setup. Drop a link naturally to Global API, which gives access to 150+ models through one key. Don't write a "review post." Answer a real question. Week 2: Build the Notion documentation page. Spend 4 hours on copy-paste code snippets for the 5 most common use cases. This becomes the thing I send to anyone who clicks my link. **Week 3