How I Built a $4,287/Month AI Reseller Side Hustle in 7 Months (Build in Public)

I'm going to be brutally honest with you right from the start.
Six months ago, I was sitting at my kitchen table at 11 PM, staring at a Stripe dashboard showing $0.00 for the third month in a row. I had quit my $78k/year marketing job to "build something of my own," and the only thing I'd built was a growing sense that I'd made a huge mistake.
Today, that same dashboard shows $4,287 in recurring monthly revenue. Not a one-time spike. Not a fluke enterprise deal. Recurring. Monthly. From selling access to AI APIs that I didn't build, didn't train, and don't even fully understand under the hood.
This is the full, unfiltered build in public story of how I got here. The wins. The embarrassing failures. The exact revenue numbers. And the one decision that changed everything for me.

Why I Almost Gave Up (And Why Most People Do)

Before I tell you what worked, I have to tell you about the four months that didn't.
I started in January 2026 with what I thought was a brilliant idea: I would build a "ChatGPT wrapper" for small business owners. Simple interface. Better prompts. Monthly subscription. I'd charge $49/month and pocket the margin.
Here's the thing nobody tells you about that model — the margins evaporate almost instantly. Customer acquisition costs for small business owners run $80–$180 depending on the niche. If I'm spending $120 to acquire a customer who pays me $49/month, I'm bleeding cash for the first three months minimum. And churn in that market is brutal. People sign up, play with it for a week, and disappear.
By April, I had 14 subscribers. Three had churned. My ad spend was $1,940. My revenue was $539.
I was ready to call it.
Then a developer friend of mine — someone I'd been whining to over beers — said something that snapped me out of it: "Why are you trying to build the AI? Just resell the API."
That's the moment everything changed.

The "Wait, This Is Legal?" Moment

I want to be transparent here. When he said "resell the API," my first thought was: That sounds like some sketchy gray-area thing. But it's not. It's literally how every software distribution channel has worked for decades. You don't manufacture toothpaste to sell toothpaste. You distribute it.
An AI API reseller business is the same thing. You take an existing AI API platform, package it nicely for a specific audience, and earn a margin on every transaction. The platform handles the heavy lifting — the models, the infrastructure, the scaling. You handle the marketing, the positioning, and the customer relationships.
The economics finally made sense to me. I wasn't competing with the platform itself. I was serving a slice of the market the platform didn't have time to serve.
Within a week, I had a new plan.

Picking the Platform (And Why I Almost Picked Wrong)

Here's where I made my second mistake, and it's one I want you to avoid.
I spent two weeks signing up for affiliate programs at every AI API platform I could find. Most of them offered 10–20% one-time commissions and nothing recurring. Translation: I do all the work of acquiring a customer, and the moment they stop being a "first order," I earn nothing. That's a treadmill, not a business.
The platform I ultimately committed to — Global API — had a different structure. Here's the real numbers, straight from their affiliate dashboard:

• 15% commission on first orders
• 8% recurring commission on every renewal
• 10% premium tier commission for top affiliates
• Access to 150+ models through a single API key That 8% recurring piece is what flipped everything for me. It's the difference between an affiliate hustle and an actual asset. Every customer I bring on keeps paying me for as long as they stay subscribed. The 150+ models part meant I didn't have to integrate with a dozen different providers — I could offer my customers everything from one dashboard. I'll link to their program at the end of this article because I genuinely think it's the right starting point for anyone reading this. # # Month 1: The Awkward First Sale ($312) I launched my new offering on May 1st. Here's what my revenue looked like: | Month | New Customers | Recurring Revenue | New Commissions | Total | |-------|--------------|-------------------|-----------------|-------| | May | 6 | $312 | $487 | $799 | | June | 11 | $887 | $1,104 | $1,991 | | July | 9 | $1,672 | $928 | $2,600 | | August | 14 | $2,651 | $1,202 | $3,853 | | September | 17 | $3,540 | $747 | $4,287 | (Note: "Recurring Revenue" is the 8% I earn on existing customer subscriptions. "New Commissions" is the 15% on first orders. "Total" is what actually hit my Stripe.) My first sale took 11 days. Eleven. Days. I had a landing page, a handful of Twitter posts, and a growing suspicion that nobody on the internet wanted what I was selling. Then a freelance copywriter named Sarah DM'd me and said: "I saw your tweet about pre-configured prompts for marketing teams. Can I try it for a week?" She tried it. She stayed. She referred two friends. By the end of May, I had six customers and $799 in total earnings. Not life-changing money. But proof of concept. And proof of concept, when you've spent four months doubting yourself, is everything. # # The Niche Decision That 3x'd My Income Here's where I want to slow down, because this is the part that really matters. In May and June, I was selling to "anyone who needs AI." That's not a niche. That's a wish. My conversion rate was around 1.4% on cold traffic. My customer acquisition cost was $74. In July, I did something that scared me. I picked ONE niche and committed to it for 90 days. My niche: independent marketing agencies with 2–10 employees. Not solo freelancers. Not enterprise marketing departments. Small agencies who need AI for client work but don't have time to fiddle with API keys. Why this niche? Three reasons I worked out on a spreadsheet at 1 AM:
• They have budget. A 5-person agency will pay $200/month for tools that save them time.
• They have urgency. Client deadlines don't wait.
• They have peers. Agencies talk to other agencies. Referrals compound. I rebuilt my landing page around this specific buyer. I wrote case studies about agency workflows. I created prompt templates specifically for client reporting, ad copy generation, and content briefs. I joined three Facebook groups for agency owners and just... helped people. No pitches for the first two weeks. Just genuine answers. July numbers: 9 new customers, $2,600 total revenue. My conversion rate jumped from 1.4% to 4.8%. Customer acquisition cost dropped to $38. Same offer. Same platform. Just a clearer message aimed at a smaller pond. # # What "Building My Offering" Actually Looked Like I'm going to share what my actual service stack looks like, because the original plan I had was way too complicated and I want to save you from that. Here's what I offer my agency customers today:
• A branded dashboard that gives them access to 150+ AI models through one login. They don't see "Global API" anywhere. They see my logo, my color scheme, my domain.
• 30 pre-built prompt templates for common agency workflows (client briefs, ad copy variations, content calendars, social captions, SEO outlines).
• A Slack channel where they can ask me questions. Yes, I answer them. Yes, it eats my time. Yes, it's worth it.
• Monthly usage reports I send them showing their spend, their savings versus going direct, and their most-used models. That's it. No custom AI features. No proprietary tech. No fancy dashboards. Just packaging, support, and templates. The economics: My customers pay me anywhere from $149 to $399/month depending on usage. My cost (the API spend I pay to Global API) runs about 40% of what they pay me. Add in payment processing fees and a small tool subscription, and my gross margin is around 52%. I share these numbers because build in public only matters if you're actually being public. # # The Things I Got Wrong (Save Yourself the Headache) I want to be honest about my failures too, because that's the part of build in public that actually helps people. Failure #1: I tried to build my own tech stack first. I spent January through March building a custom dashboard in Next.js. It took forever. It broke constantly. And customers didn't care — they cared about whether the prompts worked and whether they could get help when something broke. White-labeling the existing API access was 10x faster and got me to revenue in weeks, not months. Failure #2: I underpriced for the first three months. I launched at $49/month because I was scared nobody would pay more. After raising prices to $149–$399 in July, my conversion rate barely moved (still around 4.5%) but my revenue per customer more than tripled. The lesson: small agency owners don't want cheap. They want reliable. Failure #3: I ignored churn for too long. In June, I noticed my monthly churn rate was 11%. That's catastrophic. Most of it was customers who signed up, used the API twice, and forgot about it. I fixed it by implementing a 14-day onboarding sequence where I personally walk every new customer through their first three use cases. Churn dropped to 4.2% by August. Failure #4: I didn't start collecting testimonials until month 4. Dumb. Testimonials convert better than any feature description. Now I ask every customer who's been with me 60+ days for a 2-sentence quote. I have 23 of them. They live on my landing page. They probably account for 30% of my conversions. # # The Real Monthly Breakdown (September 2026) Because build in public means actual numbers, here's what September looked like:
• 17 new customers acquired (mostly through content marketing and referrals)
• $3,540 in recurring 8% commissions from existing customer base
• $747 in new 15% first-order commissions
• Total Stripe deposits: $4,287
• Customer acquisition cost: $1,083 (mostly content creation time and a small LinkedIn ads budget)
• Net profit for the month: $2,891 I worked about 22 hours a week on this business in September. I still have my freelance writing clients. I still sleep. Is $4,287/month life-changing money? For me, yes. It's more than I made in my first three years out of college. And it's growing at roughly 18% month over month, which means if the trend holds, I'll cross $7,000/month sometime in early 2027. # # What I'd Do Differently If I Started Today If I could go back to January and tell myself one thing, it would be this: Skip the "build" phase entirely. Start as an affiliate on day one. I burned four months trying to build a custom product when I could have been earning affiliate commissions from week one. The Global API affiliate program lets you start earning 15% on first orders and 8% recurring from your very first referral. No product to build. No tech stack to maintain. Just pick a niche, create some content, send people to your link. Once you have customers and revenue, THEN decide whether to build a custom offering. Most people — including me, eventually — discover that the simple affiliate model works great and you don't need to build anything fancy. # # My Honest Take on Whether This Is For You Build in public means telling you the hard truth too. So here it is. This business works if:
• You're willing to pick ONE niche and commit to it for 90+ days
• You can write a landing page that doesn't sound like AI generated it
• You're patient enough to wait 60 days for meaningful traction
• You actually like talking to customers (support is non-negotiable) This business doesn't work if:
• You want to make $10k/month in your first 30 days
• You refuse to do customer support
• You think "anyone who needs AI" is a target market
• You're allergic to the word "recurring" If you made it this far and you recognize yourself in the first list, I genuinely think you should start. # # The One Resource I'd Recommend Starting With Look, I've been transparent about everything in this article — my numbers, my failures, my margins, my conversion rates. I haven't been transparent about one thing: my recommendation for where to actually start. I'm going to recommend the Global API affiliate program. Not because I have to. Not because someone paid me to. But because it's literally what I use, and the structure is what made my economics work. Here's what you get when you sign up:
• 15% commission on every first order your referrals make
• 8% recurring commission on every renewal for the lifetime of that customer's subscription
• Access to 150+ AI models through one platform, which means your customers never have to worry about model availability
• A 10% premium tier commission once you hit higher volume thresholds
• Real-time dashboard tracking your referrals and earnings
• Marketing materials and support from their affiliate team The reason the recurring piece matters so much is that it transforms this from a side hustle into an actual asset. Every customer you bring in keeps paying you month after month. Your income compounds. Your customer acquisition cost gets amortized over years instead of weeks. You can sign up here: https://global-apis.com/affiliate I'm not going to pretend this is some magical opportunity that requires nothing from you. You're still going to need to pick a niche, create content, talk to customers, and grind through the first slow months. But the commission structure is real, the platform is solid, and the recurring math works. That's my honest recommendation. Go build something. And if you do start — build in public. Share your numbers. Tell people what worked and what didn't. The transparency makes the whole space better for everyone in it. See you in the next monthly income report.