Why I Built an AI That Runs My Affiliate Site — And What I Learned
Photo: Dell / Unsplash
At some point in late 2025, I decided to build an AI system that I absolutely did not need.
I'm a product manager at a hardware company. I spend my days thinking about how technology gets built, packaged, and put in front of people who might actually use it. I read about AI constantly. I sit in meetings about it. I have opinions about it that I mostly keep to myself because opinions without experience are just vibes dressed up as expertise.
That's actually what got me. The gap between knowing about something and knowing it.
I could explain roughly how large language models work. I could tell you why context windows matter, why inference speed is a competitive advantage, why the race to compress model size without sacrificing quality is one of the more interesting engineering problems of the decade. But I couldn't tell you what it actually felt like to build something with these tools. To hit the wall. To figure out why it wasn't working. To get it working anyway.
So I started building. Nights and weekends. A visual canvas where you connect AI nodes — a prompt builder here, a model call there, some logic in between. Something that could run automated workflows without me being in the room.
I called it Flux.
What Flux actually is
Flux is an AI automation system that runs on a Mac mini M4 Pro sitting on my desk. It's a visual canvas — you drag nodes onto a board, connect them, and the result is a workflow that runs on its own. One node fetches news. Another writes a summary. Another formats it for the web and publishes it. The whole chain runs at 7AM without me touching anything.
Flux runs informitIV. Everything you read here — The Scoop stories every morning, the product research, the curation — goes through that canvas. I built the tool, and then I built this site to give the tool something real to do.
Three things I learned building it
1. There is no substitute for actually doing it
Reading about AI is useful the same way reading about swimming is useful. You'll have the vocabulary, you'll understand the theory, and then you'll get in the water and immediately realize you don't know anything. The errors I hit building Flux taught me more about how these models actually behave than any article I'd ever read. They have real failure modes. Real latency. Real opinions about what they will and won't do depending on how you ask. You only find that out by asking.
2. What one person can build right now is genuinely absurd
I built a system that reads the news, writes summaries, fetches relevant product images from the internet, matches them to affiliate links, and publishes a formatted web page — automatically, daily, on hardware that cost less than a used car. Five years ago this would have been a small engineering team. Three years ago it would have been expensive and fragile. Now it runs on a desk in Minnesota and costs me about as much per month as a streaming subscription. That's not normal. That's a genuinely strange moment in history, and most people haven't fully clocked it yet.
3. Most people think this isn't for them. It is.
The barrier to using most of this technology is lower than it looks from the outside. A lot lower. The tools are getting easier. The hardware is getting cheaper. The documentation is getting better. What's still missing — and this is why informitIV exists — is someone translating it into plain language. Not dumbing it down. Just not assuming you already know what a context window is. The technology is ready for regular people. The communication hasn't caught up yet.
What this site is
informitIV is the output of all of that. It's where I share what I find.
Products that actually deliver — not because a brand paid me to say so, but because I looked at what works at each price point and wrote it down. Tech worth paying attention to — because I'm in this stuff all day and I know which news matters and which is noise. Honest takes, written by a real person who has made real mistakes and is still figuring things out.
It's not a faceless affiliate farm. It's not an AI blog that pretends to have opinions it doesn't. It's one guy sharing what he's learning, in real time, updated every morning.
This is just the first post
Every morning, The Scoop gets updated with four stories worth reading. The goal is to make this the one tab you open before anything else. Every week, I write something here — something longer, something with more context, something that tries to connect the dots between what's happening in AI and what it actually means for the way you live and work.
We're at an interesting moment. The technology is moving fast, but the part that matters — the part where it becomes useful to regular people, not just researchers and engineers — that part is just getting started.
I built something to help figure that out. This is where I'm sharing what I find.
Come back tomorrow. Something new drops at 7.
Written by
Joey
Product manager. Builder. informitIV.io