From $150 domain to $2M/year AI SaaS as a self-taught coder

Executive overview

Most founders spend months validating ideas. Jozef Gherman bought a $150 domain, reverse-engineered an AI detector in a few days, and had paying customers within a month. StealthGPT — a tool that makes AI-generated text undetectable — grew to $190,000 MRR in 15 months.

The insight: find a problem created by a new technology trend, move fast, and pour early revenue straight back into the one marketing channel that converts.

Marketing distribution matters as much as the product itself — without eyeballs, nothing else works.

From failure to idea

• Previous venture (J Squared Invest) ended with a $400,000 CFTC settlement after crypto markets collapsed in 2018
• Learned to code through Harvard CS50 and YouTube tutorials — no formal background
• Saw ChatGPT rise in 2022 and anticipated the counter-trend: demand for AI detection
• A TikTok video about GPT-0 detector confirmed the market existed
• Spent a few days reverse-engineering the detector until every output scored as "human"
• Bought the domain in February 2023 ($150); first customers arrived in March

Building and launching the MVP

• Built on Next.js and React; deployed via Vercel for fast rollbacks
• Listed on Futurepedia with 10 free trials before requiring a subscription
• First customers arrived via a TikTok creator (Phil) who covered new Futurepedia listings
• Three pricing tiers: $4.99, mid-tier, and $15/month
• Key infrastructure: Supabase (database), Vercel (deployment), Discord (team comms), Zendesk (support)

Growth and marketing strategy

• Early stage: UGC-style TikTok and Instagram content focused on market education — explaining what the product does
• Scaled aggressively with Google Ads once past summer 2023; cost per conversion dropped to $10–$20
• Reinvested almost all revenue into ads — spending ~$100,000/month on marketing
• Hired entirely through Twitter posts; no pedigree required

Business metrics (at time of filming)

• $190,000 MRR; ~8,500 active subscribers out of 400,000 total signups
• Net profit: $10,000–$30,000/month after expenses
• Monthly costs: ~$50,000–$60,000 salaries (6–7 staff), ~$5,000 infrastructure
• Team hired via Twitter; no prior track records needed

Product philosophy

• Solve as broad a problem as possible — resist niche feature requests
• Use customer feedback only for quality and UX, not for roadmap direction
• Default to simplicity: one-click solutions wherever possible
• Avoid feature bloat — engineers want complexity; users want their problem solved

Advice for aspiring founders

• Start from problems you personally find frustrating or wasteful
• Pressure-test the idea honestly: will people actually pay for this?
• There has never been a cheaper time to ship — free tiers from Vercel, Supabase, and AI tools remove most barriers
• Learn continuously: follow indie hackers and founders on Twitter for real-time ideas
• Distribution is the constraint, not the product