Jenni AI: From $2K MRR to $6M ARR on 12,000 TikToks a Month — Examining the “Creator Factory” Marketing

Yen conversions in this article use a rough $1 = ¥150.

A product once mocked as “a GPT wrapper” reached $6M (roughly ¥900 million) in annual revenue purely through its acquisition machine. Jenni AI is a case that’s hard to avoid when discussing SaaS marketing in the short-form video era. But this company’s journey can’t be summed up as “went viral on TikTok.” The full picture only emerges once you include the years stuck at $2,000 MRR, multiple pivots, and the organizational design that turned “virality from luck into statistics.”

The numbers, on a timeline

Funding was minimal (about $850K total, including a $100K (roughly ¥15 million) seed check from Jason Calacanis) and the company is profitable. The team is around 9 people, and pricing is freemium: a free tier (200 words/day) plus $30/month (60% off for annual plans).

The road to rock bottom

Park’s track record was a string of failures. He started an apparel brand at 16, then a dating app; both failed. He majored in literature at UC San Diego and dropped out. Jenni AI’s predecessor was also a non-scalable business (Altum, an AI-assisted ghostwriting agency) and the early GPT-2-based product could only deliver something like a 10-20% improvement in writing speed. The stall at $2K MRR is a product of this period.

Two turning points followed. One was the pivot to SaaS enabled by GPT-3. The other was the experience of a thread by influencer Zain Kahn going viral and multiplying MRR by five in a single month. That accidental burst of virality became proof that “if you have the distribution network, the product will sell”, and from then on, Jenni stopped leaving diffusion to chance and committed to building an in-house system to manufacture it.

Inside the “creator factory”

At the core is the industrialization of UGC (user-generated-style content). At its peak, the company contracted with over 200 creators, who posted 12,000 short-form videos a month to TikTok and Reels. Its signature format, the “POV: You have an essay due” video series, has racked up 300 million cumulative views.

The compensation design is the key detail: creators earned a $15 performance bonus per paid signup, plus continuing monthly payments for as long as the acquired user stayed subscribed. The target was micro-influencers with followings in the thousands to tens of thousands. By tying pay to “signups and retention” rather than view counts, the design pointed creators’ attention toward quality customer acquisition rather than virality for its own sake. Instead of chasing a single viral hit, Jenni managed the rate of virality across thousands of attempts, at 12,000 posts a month, even a 0.1% hit rate yields 12 hits.

Why it worked

Virality shifted from luck to statistics. In contrast to Peing’s accidental virality, Jenni manufactured diffusion through hit-rate × volume. In the short-form video era, acquisition has become a contest of who can build the mass-production system, not who has the most creative single video.

There was also a perfect match between channel and customer: the student market and TikTok. Students struggling with essays are on TikTok. It’s the same principle as Tweet Hunter on X and TypingMind on X, “demonstrate where your customers already are,” pursued to its logical end.

Underneath both sat a correct diagnosis: the years at $2K MRR were a distribution problem, not a product problem. The numbers that refused to move no matter how much the product was polished moved instantly once a channel was invented. The lesson for diagnosing a stall: separate “is it the product, or is it distribution?” first.

What didn’t work, and the risks

This case tends to get consumed as a straightforward success story, but the shadow side is on record too. Before the turning point, there were three failed ventures (apparel, an app, an agency) and years of stagnation. Park kept running the company while receiving a cancer diagnosis during the growth phase, and made the heavy call of turning down an acquisition offer at age 27.

The model’s own weaknesses are also clear. (1) “Use AI for your homework”-style messaging sits right next to platform and institutional regulatory risk. (2) If UGC-style ads saturate and CPMs rise, the hit-rate × unit-price math breaks down. (3) Video content flows past and disappears. Brand doesn’t accumulate easily, so Jenni invests in a downstream process, email and in-product engagement, to retain the users the factory pulls in. The factory is the acquisition engine. Retention is the actual business.

A 12,000-Video-a-Month Numbers Game Works Because Each Video Is Cheap

Jenni’s acquisition rests on a volume calculation: even at a 0.1% hit rate, 12 hits a month come out. That calculation only works because the contractors are micro-creators with a few thousand to a few tens of thousands of followers, making short videos that cost little per piece. Swap in YouTuber sponsorships at hundreds of thousands of yen each and the same volume cannot be built on any budget. The roster of more than 200 creators was also assembled only after the company had first experienced Zain Kahn’s thread multiplying MRR fivefold in a month.

Tying pay to “signups and retention” rather than views works at any volume as a mechanism that keeps mass production from degrading into low-effort videos chasing numbers. On the other hand, the 300 million views rested on a huge segment, English-speaking students, being packed onto TikTok, and there is no basis for assuming a similar tool in Japanese could reach the same throughput. In the Japanese market this kind of UGC factory has hardly spread beyond beauty and hair-removal ads, and SaaS and apps are close to empty territory. But empty territory and a dense population of students there are two different things.

Related cases

• PhotoAI — a contrasting single-hit case built on one TikTok video adding $20K/month
• Peing — a contrast in accidental virality