How a no-code AI study app scaled from zero to $100k MRR

Executive overview

Most founders treat distribution and product as competing priorities. Julian Alvarez, co-founder of Jungle, learned by doing both — cycling through mass DMs, organic influencer pickups, paid micro-influencer campaigns, and a 400-video-per-week UGC program.

Jungle is an AI flashcard and quiz platform for students, gamified like Duolingo. It converts uploaded PDFs, slides, or YouTube videos into practice questions and has reached $80–100k MRR.

Growth hacks exploit market gaps — and every gap eventually closes. Retention is the only durable edge.

Getting the first users

• Co-founders spotted a viral tweet: "My dream AI tool would generate flashcards from anything."
• Commented publicly, then DM'd every account that engaged — got 200+ downloads and 50–100 direct messages.
• Twitter account was banned from the volume of outreach.
• Posted to AI tool directories (e.g. Future Tools); organic influencer pickup followed from novelty of the product.
• Mentor advice at launch: ignore the product for two full weeks, focus only on distribution — if nothing moves, pivot.

Micro-influencer strategy and its limits

• Targeted creators in the 5K–100K follower range ("study talk" niche) for better CPM and lower risk than macro influencers.
• Structured multi-video deals to reduce cost per video and give creators room to be creative within a guided brief.
• One medical-student creator (Agogo) produced three videos that each crossed 1M views; Jungle went from $2k to $15k MRR in two weeks, with one video alone estimated at $20k revenue.
• Attempts to replicate the formula with other creators failed; spend exceeded returns and the strategy was abandoned.
• Minimum-view clauses and more rigorous brief iteration were identified as missed levers.

UGC at scale — and why it degrades

• Moved to a UGC model: ~30 creators each posting 10–12 videos per week across their own accounts, totalling ~400 videos per week.
• Creators pose as students who organically discovered Jungle, driving ~$2 CPMs — significantly better economics than influencer deals.
• The model works because it exploits authenticity perception, but that gap is closing: every brand now runs similar UGC, feeds are saturating, and audiences are noticing.
• UGC is a growth hack; hacks are temporary by definition.

Product-led growth and retention

• ~30–40% of new users come from word of mouth — tracked directly in onboarding ("how did you hear about us?").
• The growing tree visual on the study screen increased core engagement by 70%; students in libraries ask peers about it, creating passive word-of-mouth.
• Onboarding is split into three phases to avoid overwhelming users before they experience the core value:
  1. Upload from landing page → generate questions → sign up
  2. Set exam date, enable notifications
  3. Choose character, set weekly goal
• First-time product experience is stripped of all options (no page range, no question type selection) to minimise friction to the "magic moment."
• Paywall is shown after the mini review, not before — maximising the proportion of users who see it.

Product vs. distribution

• Early-stage: earn the right to experiment by getting 30–100 new users per week minimum.
• Later-stage: both matter; the question is ratio (80/20 distribution vs. product, or inverse), not either/or.
• Retention is 10–100x harder to move than top-of-funnel; don't expect short-term results.
• Bootstrap founders can pursue a retention-first model, but it requires being cashflow positive — survival mode pushes toward immediate monetisation.
• Freemium with a well-placed paywall can outperform hard-paywall approaches when retention is strong.

Founder mindset on growth

• Neither co-founder had a growth background; they built intuition by making content themselves and studying what worked.
• Growth is nonlinear — major inflection points often arrive unexpectedly after sustained experimentation.
• The core founder skill is resourcefulness: figure it out, get hands dirty, iterate when something stops working.