Launching today
AAJ Marketing Skills
39 marketing skills for AI agents. 24 run real engines.
3 followers
39 marketing skills for AI agents. 24 run real engines.
3 followers
Ask an AI agent about your unit economics, and you get a definition. These compute the numbers. 24 of the 39 ship with runnable engines — unit economics, pipeline forecasts, churn and NRR, budget allocation, AI-citation baselines. Your numbers in, a benchmarked verdict out, arithmetic shown. The other 15 are structured methods: positioning, messaging, content briefs, win/loss, campaign sequencing — written for an agent to execute, not for a human to skim.
Hi Product Hunt — Saroj here.
I build marketing systems for Seed-to-Series B companies. Earlier this year I kept hitting
the same wall: I'd ask an AI agent a unit economics question and get a beautiful paragraph
explaining what LTV: CAC means, instead of the number for my actual business.
So I built the engines. 39 skills covering positioning, pipeline, retention, paid media
and SEO/GEO — and 24 of them ship with runnable Node code. You give them your numbers,
they compute a verdict, and they show the arithmetic.
Here's one, verbatim. ARPA $400, 78% gross margin, 2.5% monthly churn, $4,800 CAC:
LTV $12,480 · LTV:CAC 2.6:1 · CAC payback 15.4 months
▼ Below the 3:1 floor — acquisition is inefficient; lift LTV or cut CAC before scaling.
▼ Payback exceeds the ~12-month guideline — cash is tied up longer; watch burn.
Not a definition. A number, a benchmark, and what to do next. Same inputs, same answer,
every time — which is something a prompt can't promise you.
A few of them are deliberately opinionated. The citation tracker won't report movement
that sits inside run-to-run variance, so it will sometimes tell you nothing happened
where a dashboard would draw you a trend line. The campaign orchestrator refuses to write
ads until it knows what's broken. That's the point of them.
No signup, no telemetry. One command:
npx skills add sarojkjha/aaj-marketing-skills --skill unit-economics --yes
I've run these against my own work and a set of other companies, and I'd like to see them
meet numbers I haven't seen. Run one on yours and tell me what it says — especially if you
disagree with a benchmark or think a verdict is wrong. That's the feedback that makes them
better, and I'll act on it.
I'm here all day.