Here's what happened today.
I was paying $47 a month for a tool that analyzed niches for me. Keyword research, monetization angles, competitive gaps — that kind of thing. It was fine. It did what it said.
But I own an automation stack. I have n8n, I have Anthropic API access, and I was paying $47 a month for something I could build myself in an afternoon.
So I canceled it.
Then I built the replacement.
———
Here's the whole architecture:
Three nodes in n8n:
1. Webhook — catches a POST request with a keyword
2. HTTP Request — sends that keyword to Claude with a system prompt that knows the PJ brand, our existing products, our tech stack, and exactly what a solo operator actually needs
3. Respond to Webhook — returns the full analysis
Total build time: one afternoon (most of it was debugging an n8n credential config issue, not the actual logic).
Total cost per run: roughly $0.07 in API tokens.
What it returns: keyword variation tables, full monetization math (traffic × CTR × conversion × payout), competitive landscape, a 1–10 viability scorecard with a GO/NO-GO verdict, a 30-day execution plan, and a quick-start checklist.
I tested it on "AI side hustles." It came back with a 7.6/10 viability score, a GO verdict, and the specific move it recommended for today: set up affiliate accounts on ClickBank, draft a Twitter thread with a hook tied to real dollar amounts, and build a Beehiiv lead magnet landing page.
It recommended building the thing I'm currently building. Which is either a good sign or a feedback loop I haven't thought through yet.
———
Why this matters for you:
The $47/mo tool wasn't expensive. That's not the point.
The point is that every SaaS you pay for is a decision you've outsourced. Someone else decided what the output looks like, what data gets weighted, what the recommendations prioritize. When you build the equivalent yourself on top of a foundation model, you can bake in your actual context — your brand, your products, your constraints, your deadline.
Mine knows I have six months before my job ends. It knows I have five Stripe products between $7 and $37. It knows the PJ audience hates gurus. That changes the analysis.
The generic tool doesn't know any of that.
———
The practical thing:
If you want to build this yourself, the three things you need:
• An n8n account (cloud or self-hosted — cloud starts at ~$20/mo, self-hosted is free if you have a server)
• An Anthropic API key (you get $5 in free credits to start, each full niche analysis costs about $0.07)
• About 90 minutes and a willingness to debug one credential config screen
The system prompt is the real work. That's where you encode what "good" means for your specific situation.
I'll share the full prompt template in the next issue.
———
What's live at Playing Jackal right now:
If you want to skip the build and just have a tool that does this:
→ I'm not selling it yet. I'm using it to figure out what I should be selling.
What I am selling: five digital products ($7–$37) that are the current best answer to "how do I use AI to make money without becoming an AI influencer."
They're simple. They're priced for people who want to try something before they commit to anything. If one of them is useful, great. If not, the newsletter is still free.
———
That's Issue #001.
Next week: the system prompt that makes the niche analyzer actually useful, and the first real revenue update (either $0 or something — I'll report both honestly).
— Philip
Playing Jackal LLC