How a 17-Year-Old Built a $1M/Month App With TikTok Creators

Zach Yadgari was 17 years old when he built Cal AI, a calorie tracking app that uses your phone camera to estimate the calories in any meal. Four months later, the app was generating $1 million per month in revenue. No venture capital. No paid ads. Just TikTok creators.

Cal AI went on to generate over $40M in total sales before being acquired by MyFitnessPal. Here's how Yadgari built the creator machine behind it.

The Product-Content Fit

Cal AI had one massive advantage: the product was a natural TikTok video.

Point your phone at a plate of food. The app analyzes the image and tells you the calories. That's a 6-second video that's inherently interesting, surprising, and shareable.

This is what makes TikTok creator programs work or fail. If your product can be demonstrated in a video, creators can show (not tell) its value. Cal AI's demo was so visual that even bad videos generated curiosity.

The Creator Strategy: From DMs to 150 Influencers

Yadgari didn't use an agency or a creator marketplace. He found creators the old-fashioned way.

Step 1: Curate the feed. He intentionally followed only health, fitness, and wellness creators on TikTok. This trained the algorithm to show him the exact people he wanted to work with.

Step 2: DM directly. He messaged creators with a simple pitch: try the app, make a video, get paid. No complex briefs, no brand guidelines. Just the product and creative freedom.

Step 3: Scale with virtual assistants. As demand grew, he hired VAs to manage creator outreach at scale, using the same direct DM approach.

By month 7, Cal AI had 150 influencers posting regularly.

The Economics: $5 CPM Benchmark

Cal AI's cost discipline was built around one number: $5 CPM (cost per thousand views).

The formula was simple. If a creator averages 100K views per video, they're worth $500 per video. If they average 50K views, they're worth $250. The benchmark stayed consistent regardless of follower count.

This approach rewarded engagement over audience size. A creator with 30K followers and consistently high views was more valuable than one with 500K followers and declining engagement.

Cost reduction tactics:
- Multi-video bundles (4 videos/month at a discounted rate)
- Monthly retainers for top performers
- Performance bonuses for videos that exceeded view benchmarks
- Focus exclusively on micro-creators (10K-100K followers) where rates are lowest

Why 150 Creators Requires a Tracking System

Here's the operational reality of managing 150 creators: that's 150+ videos per week. At minimum.

Each video has its own view count, engagement rate, like count, comment count, and share count. To calculate whether a creator is hitting the $5 CPM benchmark, you need to track every post.

At 5 creators, you can do this manually. At 15, it takes an hour per week. At 150, it's impossible without a system.

Cal AI needed to answer these questions weekly:

• Which creators are above the $5 CPM target? (Keep and scale)
• Which are below? (Coach, renegotiate, or cut)
• Which videos outperformed? (What format should we replicate?)
• Which creators are posting on schedule? (Consistency matters for the algorithm)
• What's the total spend vs. total views this month? (Is the program ROI-positive?)

Without per-post analytics across the entire roster, none of these questions have fast answers.

The Scaling Playbook

Cal AI's growth followed a clear progression:

Month 1-2: 5-10 creators, testing product-content fit, learning which hooks work.

Month 3-4: 20-50 creators, winning formats identified (reaction videos, "is this accurate?" tests, meal prep content), $1M/month revenue hit.

Month 5-7: 50-150 creators, VAs managing outreach, retainer deals with top performers, continued growth to $1.14M MRR.

Beyond: Continued scaling led to $40M+ in total sales and acquisition by MyFitnessPal.

What Founders Can Learn

Start with the $5 CPM benchmark. It forces discipline. If a creator doesn't deliver $5 CPM or better after 3-4 videos, move on.

Focus on one niche. Cal AI only worked with fitness and wellness creators. This kept the audience relevant and the conversion rates high.

Give creators freedom. Yadgari didn't send scripts. He sent the app. The best-performing content came from creators who genuinely enjoyed using the product and found their own creative angle.

Track per-post, not per-creator. A creator might have 3 average videos and 1 that generates 500K views. If you only look at averages, you miss the format that worked. Per-post analytics let you identify winning content, not just winning creators.

Scale gradually. 5 creators first, then 15, then 50, then 150. Each phase taught Cal AI something about which creators, formats, and economics worked best.

If you're building a creator program and need to track per-post performance across your roster, rostr handles this automatically. Paste your creators' usernames, get their analytics, see who's hitting your benchmarks. It's the tracking layer that lets you scale from 5 creators to 150 without losing visibility.