SAASPOCALYPSEverdict #SAASPOCALYPSE-3FE0
scanned 2026.04.27 · 18:03
subject of investigation
saaspocalypse.dev
▸ SaaS buildability verdict generator
verdict: WEEKEND
buildability score
78
/100
tier · weekend
the blunt take
“A tool that tells you whether you can build a SaaS — that you could build in a weekend. The ouroboros has logged in.”
The core is: scrape a URL, feed the HTML to an LLM, render a structured JSON verdict as a pretty report page. That's it. The snark is a prompt. The score is vibes-as-a-service. The irony is free.
cost breakdown.
their price ←→ your price
what they charge●
Free forever
$0
/ per URL
※ Optional $2 one-time build guide upsell
annual:$0
what it costs you✦
01 · Vercel (hobby tier)$0.00
02 · Supabase free tier (verdicts cache + directory)$0.00
03 · OpenAI GPT-4o API (~500 verdicts/mo @ ~$0.01 ea)$5.00
04 · Firecrawl or Jina.ai (URL scraping, free tier)$0.00
05 · Domain (saaspocalypse.dev)$2.00
TOTAL / mo7
▸ break-even:immediately — it's $0 forever, so the only cost is your Sunday
or, you know, use one of these.
if building feels spicy
option A
A Notion page + your own brain
Manually assess any SaaS yourself. Free, slower, less snarky, arguably more accurate.
option B
ChatGPT with a custom prompt
Paste the homepage text, ask 'can I build this solo?' — same output, zero infra, no URL bar.
option C
IndieHackers community post
Ask 'could I build X?' and get 40 opinions in 20 minutes. Noisier, free, surprisingly good.
what'll actually be hard.
est. total: 18 hours
▸ 3h scraping + LLM prompt engineering · 6h verdict UI (the score card, stack receipt) · 4h recent verdicts directory · 3h deploy + domain + edge caching · 2h writing the snarky copy (longest part)
easy
medium
hard
nightmare
01
easy
Structured LLM output
JSON schema + function calling. GPT-4o won't even argue with you.
02
easy
Verdict card UI
Score ring, tier badge, stack list. Tailwind + a bit of flair. One afternoon.
03
medium
URL scraping reliability
SPAs, paywalls, bot detection. Firecrawl handles most of it; the rest you 404 gracefully.
04
medium
Public verdict directory + caching
Store rendered verdicts in Supabase, serve from cache. Avoids re-querying the LLM on repeat scans.
05
hard
Prompt quality at scale
Getting the score, tier, and snark consistently right across wildly different SaaS categories takes iteration. The prompt is the product.
06
hard
Avoiding hallucinated pricing/stacks
LLMs confidently invent pricing. You need grounding heuristics or a web-search tool call to stay honest.
recommended stack
Next.js 15 + Tailwind (verdict UI + directory)OpenAI GPT-4o with structured outputs (the brain)Firecrawl or Jina.ai (URL → clean text)Supabase (verdict cache + public directory)Vercel (hobby tier, edge functions)
ready to build?
We'll email you the build guide. You'll be done by Sunday.
▸ generated with love, by a heartless robotverdict v2.1 · saaspocalypse.dev