How Jungle grows

A tweet went viral saying the author's dream AI product would turn anything they were learning into flashcards with AI. The founder commented that they were building something similar (it was only tangentially similar at the time) and chose the idea because of the response.

DMed everyone who engaged with the viral flashcards tweet, hundreds of people, inviting them to try the product. The one-on-one channel made it easy for users to send feedback.

Targeted micro to mid-tier creators in the study niche (StudyTok) with 5K-100K followers, where CPMs and ROI were better and less of a gamble than big influencers. Negotiated three- or multi-video deals to lower cost per video. Briefs explained the pain points and included examples of top videos and viral hooks, with feedback on drafts, while leaving creators freedom to fit their own style. The founders made content themselves to build the intuition to guide and vet creators. In hindsight they would have used minimum view clauses.

Recruit creators (mostly newer, less experienced ones, though strategy can vary) who each set up a separate account and post 10-12 videos a week, posing as students who discovered Jungle and are generously sharing the find. At about 30 creators that is close to 400 videos a week. The team trains creators and gives feedback on their videos.

A tree grows on screen as the user answers questions (plus a jungle river, XP, levels, unlockable trees and a chosen character). Because students study close together on campuses and in libraries, passers-by see the tree and ask what it is.

Growth from ~$15K to $80-100K/month came mostly from product: improving retention, making sharing as prominent as possible throughout the app, building viral loops, plus a fun gamified brand story. Word-of-mouth share is tracked with a 'How'd you hear about us?' question.

Visitors can upload a document or paste a URL directly on the landing page, which runs straight into generating practice questions, so the first 'magic' moment comes before signup. Framed as defining the aha moment and cutting time to it.

First-time users see a different, simpler product: no options for page ranges, minutes, question types or counts. They get a bite-size mini review ('answer your first seven questions') to reach the core value, and the full feature set is introduced only after they finish it.

A mentor's rule: after launching, don't touch the product for 2 weeks and spend all effort on distribution; if nothing happens, probably no one cares and you should pivot. Aim to get at least 30-100 new people in the door each week so experiments have enough users to read.

Posted the product on AI tool directories such as Futurepedia. Content creators picked it up and made videos on their own, helped by how novel AI-generated flashcards were at the time.

Phase 1: ask users to sign up only after they have generated questions. Phase 2: ask which exam it is for and when (students) or to turn on reminders/push notifications (mobile). Phase 3: on returning to the home screen after the review, pick a character, answer 'How'd you hear about us?' and set a weekly goal; an upsell for more questions also appears. Keeps new users from being overwhelmed before they've tried the product.

The paywall appears right after the first mini review, where about half of users get, instead of after a later step fewer reach, so more people see it. Users can tap 'Maybe later' and keep using a free tier, and the product aims to convert them later through retention.