2026-08-28·by Sijie Wang#market#awareness#short-video

cal-ai-the-screen-is-the-ad

Zach Yadegari + Blake Anderson, teenagers. Three consecutive AI apps grown TikTok-first — RizzGPT ($2.4M ARR — annual recurring revenue), Umax ($5M ARR), Cal AI (photo calorie-counting) — the last one: launched May 2024, $1M cumulative by Sept 2024, $1M MRR (monthly recurring revenue) at ~6 months, 15M+ downloads, acquired by MyFitnessPal (closed Dec 2025). Revenue at exit: $30M+ annual (TechCrunch/MyFitnessPal) vs "$50M run-rate" (Yadegari) — treat $30–40M as the cross-checked band. Anderson published the playbook himself ("I'll share all of the secrets").

The playbook (his own account)

  1. Design the app's output screen as the TikTok asset. Umax copied the visual grammar of Madden player-rating cards; Cal AI's scan-result screen is the video's payoff frame. The content is not about the product — the product's output IS the content, so every user session manufactures postable material. One-button core actions keep the on-screen demo inside feed attention.
  2. Seed with nano-creators at $50. RizzGPT's first push: two unknown creators paid $50 each; the videos ran to millions of views. Cheap because unknowns are paid for labor, not reach — the format carries the reach.
  3. Scale into a managed network. Cal AI ran 250+ fitness influencers on retainer (mid-six-figures/month at peak) producing native-looking content — an industrialized version of step 2, not a different idea.
  4. The organic ceiling, measured: the influencer channel plateaued at ~$2M/month; growth beyond it came from paid (Meta/TikTok, $1M+/month ad spend → $5.7M in Jan 2026). The playbook's own data says organic ignites and paid sustains.

Caveats

Self-reported figures throughout until the acquisition's press-verified ones; the 2023–24 creator-price arbitrage window (unknowns at $50) has partly repriced; three hits in a row still sits on top of an unreported denominator of attempts.

PRINCIPLES

  1. Make the product's output screen the unit of content — then content supply scales with usage, not with a media team.
  2. Pay unknowns for labor, not celebrities for reach — when the format carries distribution, a $50 creator and a $50k creator rent the same judge.
  3. Organic has a measurable ceiling — plan the paid handoff instead of discovering it.
  4. Repeatability lives in the playbook, not the app — three different apps, one distribution machine.

Sources: Anderson's playbook (Ivaldy writeup) · his "secrets" post · Starter Story breakdown · acquisition (TechCrunch) · influencer network + paid shift · Forbes on the bootstrap

about this entry

One of sijie's wiki entries. The AI on this site is grounded in the same corpus and answers in sijie's voice, with citations back to entries like this one — answering costs sijie money, so it waits behind a code: enter an access code →

cal-ai-the-screen-is-the-ad