| raw | raw/ai-content-phone-farms-zuhair.md |
|---|---|
| url |
TL;DR
Zuhair Lani did $100k/mo on TikTok Shop with fully AI-generated content, ran a $40k/mo agency in college, and raised $1M from a16z for Double Speed — software that owns the entire short-form stack: AI content creation + phone farms for posting + an “attention intelligence” feedback loop. Core distribution-automation material.
Key lessons
- Own both layers — generating creatives and deploying them on real devices. AI tools that only spit out MP4s have no data; owning posting closes the loop (“this prompt/model/format worked → recreate it”).
- Phone farms — custom software instruments human-like actions on real devices (emulators are detected now via fingerprinting); handles account creation, warm-up (searches your terms, screenshots feed → LLM relevance check → repost/comment), scheduling, personas.
- Tactic arbitrage across industries — sneaker/retail botting → restaurant waitlist hype (Supreme-style drops, 6,000-person list, sold out 2 months) → TikTok Shop posters. Every marketing strategy is a cycle that works ~a few months; “everything is an arbitrage.”
- AI does 95%, human 5% touch-up — don’t sell the “100% hands-off” dream. Prove a format works before automating it.
- Slideshow formula — name the pain point in slide 1, then “solutions” slides ordered with high-effort options first (weigh your meat daily) so your low-effort app looks like the obvious answer; first image AI-generated, rest from an image bank; fully-synthetic faces via Flux/Higgsfield LoRAs.
- Templates from a TikTok link — paste a winning video → tool generates a modular template → mass-create variations with small differentiating tweaks; auto-match hooks to app demos for hook-and-demo at scale.
Tools
Double Speed, ElevenLabs, Midjourney/Flux/Higgsfield, CapCut.
Related
- distribution-automation · app-market-research (templates from viral links)
- creator-content-engine · superwall-podcast