# From $3 to My First Real MRR: My 90-Day AI API Affiliate Build Log

> Source: <https://dev.to/truedeck/from-3-to-my-first-real-mrr-my-90-day-ai-api-affiliate-build-log-4n1j>
> Published: 2026-06-26 15:00:41+00:00

Hey, I'm an indie maker. I run three small SaaS products, a tiny newsletter, and I keep trying to layer new income streams on top without burning out. This is the unfiltered story of how I spent the last quarter building an affiliate revenue channel around AI APIs — what worked, what flopped, and what the actual numbers look like behind the scenes.

I want to be upfront about one thing right at the top: this isn't a "passive income" fantasy post. My first month earned me exactly three dollars. Three. That's not even enough for a decent lunch where I live. But by the end of month two, I was watching the first trickle of what would become real recurring revenue land in my account. And that's the part that excited me — not the size of the check, but the shape of it.

Let me back up and tell you where I started.

I've been shipping small software products for about four years now. None of them are huge. My biggest one does around $1,800 MRR, the second sits closer to $600, and the third is basically a hobby with maybe $50 a month. I'm fully bootstrapped. No investors, no cofounders, no team. Just me, a laptop, and an unreasonable tolerance for staring at Stripe dashboards at 11pm.

I also have a small developer blog that pulls in around 2,000 monthly visitors and a Twitter following of about 800 developers I've accumulated over the years by sharing builds and shipping in public.

When I decided to add affiliate marketing to the mix, I wasn't trying to replace my SaaS income. I wanted a stream that didn't come with a support inbox attached. Something that could grow while I slept, ideally — a compounding asset rather than another hourly trade.

I burned a long weekend researching AI API affiliate programs. Most of them were forgettable — flat one-time payouts that evaporate the moment the customer signs up. But three stood out as worth testing:
