How Zach Yadegari built and sold Cal AI for tens of millions at 18

Executive overview

Zach Yadegari built Cal AI — an AI-powered calorie-tracking app — from zero to $30M ARR in 18 months, then sold it to MyFitnessPal at 18 years old. Growth came in three phases: fitness influencers, expansion to broader creators, then performance ads combined with an affiliate program.

The acquisition made sense because Cal AI and MyFitnessPal shared a mission and overlapping users — combining forces was faster than competing.

The biggest unlock was treating paid ads creative as the entire strategy, not as a channel.

The acquisition and numbers

• Cal AI reached $30M revenue in 2025; by January 2026, monthly revenue was $5.7M ($50M+ ARR run rate)
• MyFitnessPal acquired 100% of the company; deal terms undisclosed but multiple is inferable from the ARR
• Cal AI and MyFitnessPal will remain separate apps; Zach stayed on through transition
• The deal took roughly 18 months from launch to sale

Growth phase 1: influencer marketing

• Started with fitness influencers doing UGC-style videos; scaled to ~$2M/month before hitting a ceiling
• Big fitness influencers share audiences — diminishing returns set in quickly
• Expanded to non-fitness influencers (broader reach, different audiences)
• MrBeast sponsorship cost $500K; was slightly unprofitable on direct ROAS but generated brand authority that unlocked future deals
• Attribution for influencer spend was always approximate: custom product page links, promo codes, revenue spikes on posting days, incrementality tests

Growth phase 2: performance ads

• Hired an agency first — they hit a $5K/day spend ceiling due to overcomplicated campaign structures
• Brought ads in-house, rebuilt with a simple structure: one scale campaign (CBO), one testing campaign
• No interest or lookalike targeting — rely entirely on creative to target; the algorithm handles placement
• Key attribution trick: custom product pages on the App Store, linked directly from each ad — gives clean per-ad revenue data without relying on Meta's pixel
• Optimise for "start trial" events, not conversions — Apple's privacy rules mean data sent after a 3-day trial is too stale for the algorithm to learn from
• Built an in-house affiliate creator program (using Tribe) where creators earn a revenue share — this, combined with paid ads, pushed revenue to $5.7M/month

Creative strategy

• Influencer videos embedded the app subtly in lifestyle content — low sell, relied on personal trust
• Paid ads required fully dedicated, direct-response videos: short hook + immediate app demo
• Built a separate creator version of the app with the Cal AI logo prominent on every screen and fake-perfect streaks — optimised to look great on camera
• The algorithm finds converting ad sequences itself; media buyers manage frequency and creative iteration, not audience settings
• Retargeting done through creative copy ("you've probably seen our ads before") rather than Meta audience settings

Paid ads principles

• Campaign structure is an art, not a science — keep it simple; complexity is the main agency failure mode
• Creative is the whole strategy; set up the account once, then iterate on creative
• Learn the 20% that drives 80% of results, then hire someone who knows the rest
• Spend $50–$100/day to experiment and build intuition before scaling
• Organic brand awareness compounds with paid ads: warmed-up audiences convert better

Team and talent

• Co-founder setup: CEO (Zach, vision/marketing), CTO (Henry Langmack, technical), CMO/COO (Jay Castillo, marketing operations), and an advisor/early investor (Blake Anderson, go-to-market strategy)
• Early growth was founder-led; the scaling breakthrough came from hiring specialists with external company experience
• Team grew to ~30 full-time plus contractor virtual assistants before the sale
• Key lesson: co-founders need complementary skill sets with clear decision-making ownership per domain — overlapping skills cause slowdowns

Operating principles

• Speed was the company's stated core value — identify bottlenecks in real time, resolve them in the same call
• High standards on product design: iterated until Cal AI was considered best-in-class, then kept going
• Stepped back briefly at college start; morale and growth stalled — the CEO's role in setting vision is underrated at scale
• Self-belief and language matter: speaking in definite future tense ("when I'm rich" not "if I'm rich") shapes subconscious confidence

Building apps today

• Vibe coding tools (Rourke and others) are now good enough to ship a revenue-generating app without engineering experience
• Creative and marketing skill beats technical skill at the current moment — a great app with bad marketing gets no downloads; a mediocre app with great marketing still converts
• Ideal path for solo founders: solve your own problem, validate with organic content, reinvest first revenue into UGC creators, then add paid ads to scale
• For apps with subscription models, performance ads (TikTok, Meta) have the highest ceiling; other channels may have better early ROI
• Know your marketing strategy before building — Cal AI's influencer playbook was planned before launch

College and education

• Attended college primarily for social capital, not education — met a small group of high-value people worth the time
• Switching out after the acquisition; would have left after year one regardless
• College curriculum is years out of date; needs a dedicated innovation officer to keep pace
• Memorisation should be replaced with first-principles problem solving and AI integration
• Best education Zach received: one entrepreneurship professor who had students build, register, and pitch actual companies