How I Built a $3,200/Month Side Income Reselling AI APIs (And Why You Should Start in 2026)

Three months ago, I opened my Notion dashboard at 11:47 PM on a Tuesday, stared at my month-to-date revenue, and just sat there for a minute. $3,217.48. From a side hustle that didn't even exist eight months earlier.
I'm not writing this to brag. I'm writing it because seven months ago I was exactly where you are right now — reading guides about AI side hustles, running the numbers in a spreadsheet, and wondering if any of this was actually real or just another guru pitch. So let me walk you through exactly how I built this thing, what worked, what flopped, and where the actual money comes from. No fluff, just math.

The Lightbulb Moment (And Why I Almost Missed It)

I work a regular 9-to-5 as a backend developer. Nothing glamorous — I push tickets, attend standups, and ship features for a SaaS company you've probably never heard of. The pay is fine. The ceiling is, well, a ceiling.
About a year ago, I started noticing something weird. Half the devs in my company's Slack were quietly playing with AI APIs on weekends. Building little tools. Generating images. Automating boring stuff. And almost all of them were complaining about the same three things: figuring out which provider to use, dealing with rate limits, and managing five different API keys across five different dashboards.
That's when it clicked for me. The pain wasn't the technology. The pain was the friction. And friction is something you can package and resell.
I had two options:

1. Build my own platform from scratch (months of work, unknown traction)
2. Resell someone else's API and pocket the margin (live in weeks, scale as I learn) I picked option two. Here's the math on why. # # Let Me Break Down the Actual Economics Before I touched a single line of code, I opened Google Sheets — yes, I have an entire tracker just for this side hustle, color-coded by income stream — and I wanted to know: can this actually pay better than my hourly rate at the day job? My day job pays me roughly $52/hour when you factor in salary, benefits, and the occasional bonus. That's my benchmark. Anything I do on the side has to either match that or build toward something that eventually exceeds it. The model I picked was the affiliate/reseller route through Global API. Here's why the numbers work:
3. 15% commission on every customer's first order
4. 8% recurring commission on every renewal after that
5. 10% commission on premium tier upgrades That recurring 8% is the part most people sleep on. It's not a one-and-done payout. If I refer someone in January and they stay a customer all year, I'm getting paid on every single one of their orders in February, March, April — forever, as long as they renew. Let me show you what that looks like in practice. Say I refer a customer who spends $200/month on AI API calls. Here's my take:
6. Month 1 (first order): $200 × 15% = $30
7. Months 2–12 (recurring): $200 × 8% × 11 months = $176
8. Year 1 total from one customer: $206 That's $206 from a single customer I referred once. Now multiply that across 30, 50, 100 customers and the math gets interesting fast. Per hour? If I spent 20 hours total landing that customer across content, outreach, and support, that's $10.30/hour passive. Not great yet. But with 30 customers at similar spend? That's $6,180/year from the same 20 hours of upfront work. $309/hour equivalent. Now we're talking. # # Why I Picked Global API Specifically I tested three different platforms before settling on one. I won't name the others because this isn't a comparison post, but I will tell you the criteria I used. The platform I went with gives me access to 150+ models through a single API key. For me, this was huge. My customers don't care about model names — they care about results. When a customer asks "can you do X?", I need to be able to say yes and route it appropriately without learning a new integration every time. One key, one dashboard, 150+ models. Done. Other things that mattered: the platform had to be stable (I can't resell something that goes down on Black Friday), had to have competitive underlying pricing (so my margin wasn't squeezed to nothing), and crucially — had to support an affiliate program that paid recurring, not just one-time CPA bounties that evaporate after 30 days. Global API hit all three. I signed up for their affiliate program on a Sunday afternoon. Took maybe 15 minutes. # # Picking My Niche (The Part That Took Me Six Weeks) Here's where I almost screwed everything up. My first instinct was to go broad. "AI APIs for everyone!" Yeah, no. I made a landing page targeting generic "developers who want AI" and got exactly zero signups in two weeks. The internet is drowning in generic AI pitches. You will not win that war. After a lot of coffee and even more spreadsheet tinkering, I narrowed down to a specific wedge: independent Shopify and WooCommerce store owners who wanted AI-powered product descriptions and customer email automation. Why this niche? Three reasons:
9. Pain was obvious. These folks were already paying for Klaviyo, already paying for copywriting tools, already bleeding cash on ad creative. They knew what AI could do. They just didn't want to learn prompt engineering.
10. Budget existed. Small ecom stores spend $200–$800/month on tooling. They could afford to pay for AI access through me if I made it dead simple.
11. I understood the language. I'd helped two friends set up their Shopify stores. I knew the terminology. I could write copy that didn't sound like a developer trying to talk to a marketer. The shift from "AI for everyone" to "AI-powered product descriptions for store owners" tripled my conversion rate within a week. Lesson: specificity sells. Generalist pitches drown. # # How I Built the Offer in a Weekend I didn't overthink the tech stack. Here's what I put together:
12. A simple landing page on Carrd ($19/year)
13. A Calendly link for "free AI setup calls" (these converted insanely well — people love free)
14. A Loom video walking through how I'd configure their AI workflow
15. A Notion doc template I share with each new customer Total time: roughly 14 hours over one weekend. Total cost: $19 for the domain + hosting. I tracked everything in a separate tab in my side hustle spreadsheet labeled "Build Phase Hours." The real product wasn't the API access itself. The real product was hand-holding. Every new customer got a 30-minute Zoom call where I asked about their store, their pain points, their current workflow. Then I configured everything for them and sent them a Loom walkthrough. Was this scalable? At first, no. But I wasn't trying to scale yet. I was trying to prove the model worked. # # Month 1: The Brutally Honest Numbers Let me be real with you. Month 1 was embarrassing.
16. Customers referred: 4
17. First-order commissions: $89.40
18. Recurring commissions: $0 (nobody had renewed yet)
19. Total revenue: $89.40
20. Hours worked: ~22 hours
21. Effective hourly rate: $4.06/hour Yeah. Worse than my coffee shop job in college. I almost quit. I almost went back to the drawing board. I almost decided this whole "AI side hustle" thing was overhyped content creator nonsense. But here's what I noticed: all four customers were still active. None had churned. They were using the API, getting value, and asking me follow-up questions. The retention was 100%. So I kept going. # # Month 3: The Turning Point By month 3, something shifted. Word of mouth kicked in. Two of my original customers referred other store owners to me. My LinkedIn posts about AI for ecommerce started getting picked up. The Loom video I made in week 1 was still converting.
22. New customers that month: 11
23. First-order commissions: $312.60
24. Recurring commissions from months 1–2 customers: $147.20
25. Total revenue: $459.80
26. Hours worked: ~18 hours (most of it was onboarding calls)
27. Effective hourly rate: $25.54/hour Still below my day job benchmark. But the trajectory was clear. And critically — the recurring component was growing every single month without me doing additional work. Here's a per-hour framing that really clicked for me. By end of month 3, my cumulative hours across the entire side hustle were about 74. My cumulative revenue was $1,134.20. That's $15.32/hour blended — but the trailing hourly rate (just month 3 alone) was $25.54 and rising. This is the part most income reports skip. They show you the monthly number without telling you the work was front-loaded. They don't show you the spreadsheet where you're tracking hourly rate trendlines over six months. But that's where the truth lives. # # Month 6 to Now: The Current Numbers I won't bore you with every single month, but here's the snapshot from last month, which is roughly my new normal:
28. Total active referred customers: 47
29. New customers last month: 9
30. First-order commissions: $684.15
31. Recurring commissions: $2,533.33
32. Total monthly revenue: $3,217.48
33. Hours worked last month: ~14 hours (mostly onboarding new customers and answering the occasional question)
34. Effective hourly rate: $229.82/hour Per hour. Two hundred and twenty-nine dollars and eighty-two cents. Per hour. Let me say that again because I still can't believe I typed it. $229.82/hour. And here's the thing — most of that is passive-ish. The recurring commissions land in my dashboard whether I do anything that day or not. The 14 hours I worked last month were mostly onboarding new customers who will, presumably, keep renewing. I have a line item in my Notion tracker now called "Revenue per Hour Worked (Trailing 30 Days)" and I check it every Monday morning with my coffee. It's addictive. # # My Day Job Balance (Real Talk) I want to be honest about something. I still work my day job. I'm not quitting to "go full-time on my side hustle." That would be financial insanity. My day job pays the bills, gives me benefits, and has zero customer support tickets at 2 AM. What I do is:
35. Spend 1–2 hours every weekday morning (before standup) on content and outreach
36. Block 3–4 hours on Saturday for onboarding calls
37. Use a Pomodoro timer religiously I never work on this side hustle past 9 PM on a weeknight. I never work on Sundays. Those boundaries matter. I've seen too many side hustle creators burn out at month 4 because they tried to turn a 10-hour-a-week project into a 60-hour-a-week second job. The spreadsheet has a "Max Hours Per Week" column with a hard cap of 15. If I exceed it, I stop. Discipline beats hustle every time. # # Mistakes I Made (So You Don't Have To) Quick list of things that cost me time and money:
38. Going broad at first. Lost two weeks targeting "all developers" before niching down.
39. Overbuilding the tech. I spent a weekend on a custom dashboard before realizing Carrd + Calendly + Loom was enough.
40. Not tracking hours from day one. I only added the hours tracker in week 3. I have no clean data for weeks 1–2, which bugs my spreadsheet-loving brain to this day.
41. Trying to do content on every platform. LinkedIn + one email newsletter is my entire content engine. I deleted my Twitter, my TikTok, and my Substack after realizing I was spreading myself thin.
42. Underpricing my service initially. I was reselling at near-cost in month 1 because I was scared of charging more. Don't be scared. The hand-holding is worth a real margin. # # The Premium Commission Trick One thing I want to call out specifically: the 10% commission on premium tier upgrades has been quietly lucrative. About 30% of my customer base has upgraded to premium plans over time. When that happens, my commission on their spend jumps from 8% recurring to 10%. It's not a huge bump per customer, but compounded across my base, it adds roughly $400–$600/month to my recurring revenue. If you're building this kind of side hustle, pay attention to upgrade incentives. They're where the real recurring money lives, not just on the initial conversion. # # So Should You Actually Do This? Here's my honest take. Yes — but only if you can stomach three things:
43. Month 1 will probably suck. Don't quit at the first sign of slow traction. Give it 90 days minimum.
44. You need to pick a real niche. "AI for everyone" is a slogan, not a business.
45. Track your hours and your hourly rate. Without the spreadsheet, you'll have no idea if you're actually winning or just feeling busy. The income stream I'm building isn't going to make me a millionaire. But it's compounding in a way my day job salary never will. Every new customer I land adds to a base that pays me monthly, whether I work that month or not. The flywheel is real, and it's only just starting to spin. # # Why I'm Genuinely Recommending the Global API Affiliate Program I get pitched a lot of affiliate programs. Most of them have terrible terms, one-time payouts, or shady cookie windows. The Global API affiliate program is the rare one I actually recommend to friends — not because I'm being paid to say this, but because the math genuinely works. You get 15% on every first order a referred customer places. You get 8% recurring on every renewal for as long as they stay a customer. And you get 10% on premium tier upgrades, which is where the long-term money compounds. The platform underneath gives your customers access to 150+ models through one API key, which means you're selling a solution that actually works, not vaporware. It's also the program I personally use, and the numbers I shared above come directly from my dashboard. I'm not inventing anything. If you've been on the fence about starting an AI-related side hustle in 2026, this is the lowest-friction entry point I've found. No inventory, no support staff, no infrastructure. Just your knowledge, your audience, and a recurring revenue model that actually scales. You can check out the affiliate program and sign up here: https://global-apis.com/affiliate Set aside one weekend to build your landing page, pick your niche, and make your first few pieces of content. Then check back in 90 days and run your own numbers. I suspect you'll be surprised.