<p>AI exam prep from YouTube lectures. Students paste a YouTube lecture and get a full 10–20 page study notebook: every concept, definition, formula (LaTeX) and worked example, in a handwritten-notebook look. They also get practice quizzes with verified answer keys, flashcards, chat with the lecture and PDF export. Free tools (text / PDF / photo → quiz and flashcards) bring in organic traffic, much of it from ChatGPT recommending the site. The audience is mostly exam students outside the US: NEET/JEE/CBSE in India, Cambridge O Level, KPSS in Türkiye, CFA, med school.</p><p><b>Key highlights</b> (Stripe + production database, as of 4 Oct 2026)</p><ul> <li>Lifetime revenue: $57.80 (6 payments since June 2026), no refunds</li> <li>MRR: ~$20 (2 active Solver subscriptions; a third cancels at period end)</li> <li>Monthly costs: ~$11–16 (OpenAI $5–10, VPS $6); domain $4 for the first year</li> <li>Annual profit: roughly break-even. About $240 revenue run-rate vs ~$130–190 costs, so ~$50–110 per year at current MRR.</li> <li>Business margin: ~20–45% at current MRR; AI cost per notebook is a few cents, so margin grows with revenue</li> <li>Churn: 2 of 4 subscriptions ever started have cancelled (sample too small to be meaningful)</li> <li>LTV: ~$12 per paying customer so far</li> <li>CAC: $0. No paid acquisition; all growth is organic (ChatGPT citations + SEO).</li> <li>Traffic: 3,000–5,000 visitors per month, organic only</li> <li>Users: 278 registered (199 in the last 30 days); 97 have processed at least one lecture</li> <li>Signups by month: May 3 → Jun 7 → Jul 17 → Aug 46 → Sep 173 → Oct 39 in the first 4 days</li> <li>Usage: 177 notebooks generated (42 hours of lectures), 66 saved quizzes/flashcards</li> <li>Channel: 63 of 75 users who said where they found my platform named ChatGPT</li> </ul> <p><b>Team</b></p><p>Solo founder: product, engineering, design, marketing and support (Crisp chat + email). Development is heavily AI-assisted. The business runs without a team: lecture processing, emails and billing are automated.</p><p><b>Tech stack</b></p><ul> <li>App: Next.js 16 (App Router), React 19, TypeScript, Tailwind CSS</li> <li>Data: PostgreSQL + Prisma; Redis + BullMQ processing queue (transcription → analysis → notebook)</li> <li>AI: OpenAI (gpt-4o-mini for generation, gpt-4.1-mini for quiz answer-key verification), embeddings for chat over the lecture (RAG); youtube-transcript / youtubei.js for transcripts; KaTeX, Mermaid and rough.js for math and diagrams</li> <li>Auth &amp; billing: NextAuth (Google + email/password); Stripe (subscriptions, credit packs, automatic regional pricing via IP geolocation)</li> <li>Ops: single VPS (nginx, pm2, nightly database backups); Resend for email; VisitTrack + Umami analytics</li> <li>Localization: interface in English, Spanish, Portuguese and French; notes generated in 19 languages</li> </ul> <p><b>Marketing and growth</b></p><ul> <li>AI-search visibility: the free text-to-quiz page is cited by ChatGPT for "text to quiz" queries. ChatGPT is the main source of signups (63 of 75 self-reported).</li> <li>SEO: 88 blog articles, 7 programmatic quiz-tool pages, 8 exam landing pages (NEET, JEE Main, CBSE Class 12, UPSC, Cambridge O Level, CFA Level I, KPSS, AP Biology) with real sample notebooks, localized pages in ES/PT/FR, and a public notebook library (19 notebooks).</li> <li>Results: signups grew from 7 in June to 173 in September with $0 ad spend; traffic is now 3–5k visitors per month.</li> <li>Recent conversion work (shipped in the last week, not yet measured): regional pricing (−55% / −40% for India, Türkiye, Brazil etc.), a quiz → notebook upsell, a homepage repositioned around exam prep, and a mobile dashboard fix.</li> <li>Not tried yet: paid ads, YouTube/TikTok content, affiliates, school partnerships.</li> </ul> <p><b>Revenue and profit</b></p><ul> <li>Pricing: Free (1 lecture), Lite $4/month, Solver $12/month, credit packs $2–$35, regional prices from $1.80/month, 30% founder discount</li> <li>Revenue by month: Jun $2, Jul $12, Aug $15, Sep $28.80</li> <li>MRR: ~$20, ARR run-rate ~$240</li> <li>Costs: OpenAI $5–10/month (scales with usage), VPS $6/month, domain $4 first year (renewal at the registrar's standard .study price); Resend and Crisp on free tiers</li> <li>Profit: roughly break-even (~$50–110 per year at current MRR)</li> <li>Unit economics: AI cost is a few cents per notebook; a single $4 Lite subscriber covers the AI cost of dozens of free users</li> <li>Funnel: 3,000–5,000 visitors/month → ~170 signups/month (~4% visitor → signup) → ~1–2% of signups pay</li> <li>Retention / churn: 4 subscriptions ever started, 2 cancelled. Too few to measure reliably.</li> <li>CAC: $0; LTV so far: ~$12</li> </ul> <p><b>Return on investment</b></p><p>At $5,000, this is the purchase of a working product and a traffic channel, not a cash-flow multiple. Current profit does not pay back the price on its own, and buyers should value it accordingly.</p><p>The upside is in the funnel that already exists: 3–5k visitors and ~170 signups per month at $0 CAC. If signup → paid rises from ~1.5% to 4% (regional pricing and upsells are already live) at ~$8 average, that adds about $50 of new MRR every month. That is roughly $600 MRR within a year if retention holds, which would pay back $5,000 in about 12–18 months. This is a projection, not a result.</p><p>Bigger upside: cross-selling to an existing EdTech audience, or adding paid acquisition on top of a working free-tool funnel.</p><p><b>Startup assets</b></p><ul> <li>Full codebase (355 commits since May 2026) and production setup</li> <li>Domain</li> <li>278 registered users (276 emailable; marketing opt-outs respected)</li> <li>88 blog posts, 15+ programmatic and exam landing pages in 4 languages, 19 public library notebooks</li> <li>Stripe account with active subscriptions and pricing / coupon setup</li> <li>Existing ChatGPT and Google visibility for the quiz-tool pages</li> <li>Logo, brand and design system</li> <li>10 days of founder support after the sale: setup transfer, codebase walkthrough, help moving Stripe, domain and server</li> </ul> <p><b>Risks</b></p><ul> <li>Very early revenue: MRR is ~$20; the value is in the product and the traffic, not cash flow. Mitigation: priced as an asset; conversion levers are already built and live.</li> <li>Channel concentration: ChatGPT referrals bring most signups. Mitigation: diversify with SEO (88 articles and exam pages already exist) and new channels.</li> <li>YouTube transcript access could change. Mitigation: multiple transcript methods already in place; Whisper transcription is a possible fallback.</li> <li>OpenAI dependency and cost. Mitigation: the model is swappable through configuration; an OpenRouter fallback exists.</li> <li>Users are mostly in lower-income markets. Mitigation: regional pricing is live.</li> <li>Solo-founder knowledge. Mitigation: 10-day handover and a codebase walkthrough.</li> </ul> <p><b>Summary</b></p><p>This is a working, polished AI study product with real organic momentum. Signups grew ~25× in four months with zero ad spend, driven by ChatGPT recommending it, and the site now gets 3–5k visitors a month.</p><p>The hard parts are built: the lecture → notebook pipeline, quizzes with verified answer keys, multilingual SEO, Stripe billing with regional pricing, and exam landing pages. It runs for ~$15 a month.</p><p>What's missing is monetisation focus and distribution. A buyer with marketing skills or an existing EdTech audience can turn traffic that already arrives for free into recurring revenue.</p><p>✅ $58 TTM revenue</p><p>✅ 2 customers</p><p>✅ Business model: SaaS+Credits</p><p>✅ Built with Node.js, Next.js, Docker</p>