Our read on each play — worth doing, real-but-hard, or a trap —
with the claims we’ve actually scored shown beside it as evidence. The read is our
judgment; the scores are the engine’s. Small sample, said plainly.
20 plays · our read on each · 10 backed by a teardown so far · 14 claims scored · a read, not a market census
Feasible · 2
Worth doing.
Build-in-public micro-SaaSSlow, compounding, unglamorous — and the one pattern our math has actually liked so far.1 claim · 80
Autonomous money-making agentsThe agent runs fine; the profit needs a market it usually doesn't have. Autonomy is the overclaim.4 claims · 73 · 73 · 69 · 62
AI automation agencyReal demand — the AI build is the easy part; landing and keeping SMB clients is the actual job.2 claims · 69 · 0
AI UGC / avatar ad videoBrands do pay for UGC, but it's saturated and the edit time never actually shrinks.1 claim · 70
AI receptionist / voice-agent agencyThe tech is cheap and works; the wall is selling and retaining local businesses, not building it.1 claim · 0
AI wrapper appsShips in a weekend; dies on churn and zero moat the moment the base model shifts under it.1 claim · 73
Cold-outreach lead-gen serviceA first client is genuinely doable; scaling past your own hours is where most of these stall.1 claim · 82
Faceless YouTube / Shorts at scaleReal RPM money exists — beside a million identical channels. Reach is the gate, not the tool.1 claim · 70
AI newsletter / niche mediaWorks if you already had, or will patiently build, an audience. That precondition is the whole catch.not yet scored
AI product photography / listingsA clean B2B wedge — e-commerce sellers pay for conversion lift, not for vibes.not yet scored
AI voice-clone / dubbing servicesCheap to start and real B2B demand for dubbing — but a crowded race to the bottom on price.not yet scored
Local SMB AI consultingUnsexy, high-trust, real budgets. Underexposed precisely because it doesn't screenshot well.not yet scored
Vibe-coded 'app in a weekend' flipsBuilding is trivial now; distribution is the entire unsolved problem, and it didn't get easier.not yet scored
Debunked · 6
Don’t — the math or the market already closed it.
'Teach the AI hustle' coursesFeasible for the seller, rarely for the buyer. The one selling the map seldom walked the road.1 claim · 64
Amazon KDP / AI booksThe store is saturated and the terms keep tightening on AI-generated content.1 claim · 69
AI dropshippingAd costs and saturation already closed this. The median attempt is a net loss.not yet scored
AI faceless music (Suno etc.)Streaming pennies times algorithmic suppression of AI tracks equals a rounding error.not yet scored
AI print-on-demand / merchFlooded before the tools got good; cents of margin against effectively infinite supply.not yet scored
AI trading / money-botsThe most arithmetically falsifiable claim in the feed: the compounding usually breaks before a market is even involved. We score the claim, never the trade.not yet scored
How to read this
The read is RealityScore’s
editorial call on a play’s real-world viability — a judgment, from a still-small
sample, and labelled as one. The evidence is the actual feasibility score of
every claim we’ve torn down in that category (out of 100). We keep them apart on purpose:
one fraudulent claim is a scam we caught, not a verdict on the whole market.
Marks: ◯ feasible ◖ strained ● debunked.
A second axis — how loud a play is — stays off until it comes from a real
mention-volume source, not a guess.