ChatGPT Made This Broke Kid $1 Million a Month in 18 Months

Business turnaround & overall strategy

• Blake (non-engineer) went from “dead broke” (sleeping in parents’ attic) to ~9 crore INR/month (~$1M/month) in 18 months.
• The method wasn’t technical novelty—it was repeatedly building consumer AI apps that ride viral distribution.
• Core belief: Technology isn’t the main moat—distribution/audience is.

---

The “3-move” playbook Blake runs repeatedly (execution framework)

1) Choose an obsession + filter it for virality

• Start with what people already obsess over (e.g., dating, looks, health, status), not “what AI can build.”
• Use the Three-word travel test:
  • The app must be pitchable in 3 words to a stranger within 10 seconds.
  • If it can’t spread by word of mouth, it won’t spread.

2) Time the product to a “two-wave collision”

• Look for when a cultural trend meets a new AI capability.
• Examples of “wave collisions”:
  • Dating anxiety + ChatGPT → Rizz GPT
  • Looksmaxing + GPT-4 Vision → Umax

3) Ship fast with imperfect quality + make distribution part of the product

• Ship the ugly version: launch with bugs/missing features if needed to learn faster than “polishing.”
• Build shareable artifacts (e.g., designs optimized for screenshots) so marketing becomes compounding.

---

Concrete company/product case studies

1) Rizz GPT (initial breakthrough)

Problem (market pull)

• Users needed help crafting first messages on dating apps.

Product

• Screenshot a conversation → the app extracts text → ChatGPT generates replies.
• Includes:
  • “flirt options”
  • a spice meter for boldness

Execution & operational tactics

• Launched early despite major issues:
  • notifications broken
  • no reviews
  • accidentally exposed an API key
• Distribution-first move:
  • Found anonymous TikTok accounts posting dating/pickup-line content with millions of views despite low follower counts.
  • Paid $50 each (2 creators) = $100 total to embed the app into their slideshow content.

Results / KPIs mentioned

• In ~1 week: 200,000 downloads
• Monthly revenue:
  • shortly after: about $80,000/month
  • later: close to $200,000/month

Management/strategy lesson learned

• Distribution beats perfection on the internet.
• “Rizz GPT” was an experiment to prove the ability to spot waves early.

---

2) Umax (scale through shareability + viral loop)

Market

• “looksmaxing” / face optimization obsession.

Enabler

• GPT-4 Vision (image understanding via API), avoiding months of custom vision engineering.

Product mechanics

• Upload selfie → AI scores across 20+ traits
• Shows:
  • current score
  • potential score
• Produces improvement guidance (e.g., skin, hairstyle, grooming, fitness)

Why it spread (growth loop design)

• Scorecards are inherently screenshot-friendly (like exam scores/fitness stats).
• Users post screenshots → friends download → friends post their own score.
• The product is designed so every screenshot acts as free ad inventory.

Revenue / KPIs mentioned

• Month 1: ~$100,000
• Month 2: ~$200,000
• Afterwards: “half a million dollars regularly” (~$500k/month)
• Downloads: 10M+ Play Store downloads

Competitive response (moat under attack)

• Competitor clone: Looksmash AI
  • Copied Umax “top to bottom”
  • Matched Umax downloads in ~2 weeks
• Blake “won anyway”:
  • Umax eventually outgrew the clone by ~4x
  • Distribution moat > tech moat

Strategic takeaway

• When clones appear, distribution and audience determine survival more than model/product parity.

---

3) Cal AI (second-wave learning: AI + boring markets)

Strategy pivot

• Don’t hunt for “clever ideas.”
• Enter a giant existing market and use AI to make the process 10x easier.

Market

• Calorie tracking (tedious logging).

Product

• Photograph meal → estimate calories + breakdown protein/fat/carbs

Execution playbook application

• Build for behavior change via simplification.
• Avoid inventing demand—remove friction in an existing behavior.

---

Distribution system (marketing is the product)

Micro-creator acquisition playbook

• Find micro creators with:
  • 50–100k followers
  • millions of views per post
• Pay small amounts to many creators and iterate on the hook.
• DM workflow noted:
  • “DM about 100 creators”
  • expect ~10 reply and ~3 convert
  • then go all-in on the few “must-have” accounts
• Multi-touch path:
  • DM on TikTok → DM on Instagram → join Discord → message managers
  • Goal: be unignorable

Reinvestment rule (unit economics concept)

• If every 1,000 views generates more money than the cost to buy those views, you can reinvest and outpace competitors.

---

AI stack (tooling framework for solo/non-technical builders)

Layered architecture (a practical process playbook)

Layer 1: Code generation / build automation

• Examples: Claude Code, Cursor
• Approach: describe in plain English; AI writes/explains/fixes code.

Layer 2: Capability tools (what your AI “does”)

• Image understanding: Nano Banana, ChatGPT image models
• Voice: ElevenLabs
• Avatars: HeyGen

Layer 3: Distribution/build targets

• iOS apps via SwiftUI + App Store
• Alternative fast route:
  • web app + payment + one-click deployment

Layer 4: Marketing & distribution

• Not just audience building; treat it as a core system.
• Examples listed:
  • TikTok creators, X, YouTube Shorts, Instagram Reels
  • newsletters, communities, SEO

---

KPI/targets and performance signals explicitly stated

• Scale targets achieved:
  • ~$1M/month after 18 months
• Rizz GPT:
  • ~200,000 downloads in ~1 week
  • ~$80k/month, later ~$200k/month
• Umax:
  • Month 1: $100k
  • Month 2: $200k
  • “$500k/month regularly” (described as half a million)
  • 10M+ Play Store downloads
• Competitive dynamics:
  • Clone matched in ~2 weeks
  • Umax outgrew clone by ~4x
• Distribution conversion expectations:
  • 100 DMs → ~10 replies → ~3 conversions
• Unit economics concept (implied KPI):
  • Revenue per 1,000 views > cost per 1,000 views (enables reinvestment)

---

Actionable recommendations (condensed into “what to do this weekend”)

• Start with existing obsessions (dating/looks/health/status), not new tech.
• Use the Three-word travel test; if it needs 5 minutes to explain, it’s likely dead.
• Time the build to a cultural + capability overlap (trend + new AI capability release).
• Ship the ugly version to collect user feedback immediately.
• Design the product to be shareable by default (screenshots, scorecards, “postable artifacts”).
• Build your marketing pipeline with micro creators:
  • DM ~100, refine what converts, then scale winners.
• Prefer AI to simplify boring markets (e.g., Cal AI for calorie tracking) rather than trying to invent demand.

---

Note on presenters/sources

• Presenter specifically mentioned: Blake (the founder described).
• No other named presenters or external case-study authors are clearly identified in the provided subtitles.