Problem
Over 3 million students appear for NEET and JEE annually in India, with the majority studying in Tier-2 and Tier-3 cities where coaching is in Hindi, Tamil, Telugu, or Bengali — yet every major doubt-clearing platform (PhysicsWallah, Unacademy, Doubtnut) defaults to English-first interfaces and video-heavy formats that eat mobile data. A student in Patna or Coimbatore who gets stuck on a Physics derivation at 11 PM has no instant, affordable, vernacular channel for a step-by-step explanation.
Solution
Build a WhatsApp Business chatbot where a student sends a photo of any NEET/JEE question. The bot runs OCR to extract the problem, calls a Claude-class LLM with a subject-specific prompt, and returns a fully worked solution in the student's chosen language (Hindi, Tamil, or Telugu). Language preference is set once at onboarding via a two-tap UPI payment flow. V1 covers Physics and Chemistry (highest doubt-volume subjects); Biology and Maths added in month 3. No app install, no data-heavy video — just WhatsApp, which 530 million Indians already have.
Why Now
ChatGPT crossed 46 million Play Store downloads in March 2026 alone, signalling that AI-assisted learning has crossed mainstream adoption even in India's general population. Simultaneously, YC's 2026 India cohort shows edtech founders doubling down on vernacular-first, mobile-native products — validating that English-only incumbents are leaving a massive gap. WhatsApp's updated Business API now supports image messages and interactive buttons at scale, removing the last technical blocker for a photo-based doubt flow.
Target User
First 1,000 customers: Class 11/12 students in Tier-2 and Tier-3 cities (Patna, Indore, Coimbatore, Kota day-scholars) whose household income is ₹30,000-80,000/month, who already use WhatsApp for study groups, and who are too price-sensitive for ₹5,000/month coaching apps but will pay ₹299/month for instant doubt clearing. Acquired via WhatsApp study-group forwards and YouTube Shorts in regional languages.
Business Model
Monthly subscription at ₹299 (unlimited doubts, 2 subjects) or ₹499 (all subjects). Annual plan at ₹2,499 sold starting month 4. All payments via UPI/Razorpay link in WhatsApp. API cost per doubt is ≈ ₹0.80 (OCR + LLM); at 50 doubts/month per user the API cost is ₹40, leaving a ₹259 gross contribution on the base plan — an 87% gross margin. At 500 active subscribers, monthly revenue is ₹1.5 lakh; at 2,000 subscribers (achievable by month 9), ₹6 lakh/month with ~₹5.2 lakh gross profit.
Competitive Landscape
- Direct (India): Doubtnut (video-only, English UI), PhysicsWallah Doubts (app-required, video-heavy), Brainly (peer answers, slow)
- Direct (global reference): Photomath (US/EU, maths only), Mathpix (developer-focused); China's Zuoyebang (photo-first AI doubt app, 170M users) is the exact analogue
- Why we win: WhatsApp-native (zero install friction) + vernacular-first (Hindi/Tamil/Telugu on day 1) + instant text answer vs. 5-minute video — three simultaneous differentiators no incumbent currently bundles together
6-Month Plan
- Month 1 (₹1.5L): WhatsApp Business API setup, OCR pipeline (Google Vision), Claude API integration, Hindi language prompt tuning. Internal alpha with 20 students.
- Month 2 (₹1.5L): UPI/Razorpay subscription billing via WhatsApp payment link. Tamil and Telugu prompt packs. 100-student beta, free tier.
- Month 3 (₹1L): Subject expansion to Biology. Referral loop: share-a-solution viral mechanic. Target 300 paying users.
- Month 4 (₹1.5L): Annual plan launch. YouTube Shorts regional ads (₹50K/month). Target 700 paying users.
- Month 5 (₹1.5L): Accuracy audit pass with subject-matter educators (freelance). Wrong-answer flag flow. Target 1,200 users.
- Month 6 (₹1L): Dashboard for parents (progress report PDF). B2B pilot with 3 coaching institutes (white-label plan at ₹15,000/month for 100 students). Total capex: ₹8L.
Risks
- LLM accuracy on complex derivations (High likelihood × High impact): A wrong step in a NEET Chemistry derivation erodes trust fast. Mitigate with a "flag answer" button feeding a human reviewer queue; build reviewer network in month 2.
- WhatsApp Business API rate limits and policy changes (Medium likelihood × High impact): Meta can change pricing or restrict educational bots; mitigate by collecting user emails/phone for a fallback SMS/web channel from day 1.
- PhysicsWallah cloning the WhatsApp channel (Low likelihood × High impact): PW has 80M users but is video-infrastructure-heavy and unlikely to pivot fast; 12-month head start on vernacular prompt quality is the defensible gap.
Score Breakdown
Market (18/20): 3M+ annual NEET/JEE test-takers; even 0.5% penetration at ₹299/month = ₹54Cr ARR. Adjacent to 10M+ government exam aspirants (UPSC, SSC). Comfortably ₹500Cr+ TAM in 3 years, one point short of 20 because the price point is low and conversion from free WhatsApp study groups requires behaviour change.
Capital (13/15): MVP is pure API-stitching — WhatsApp Business + Google Vision + Claude + Razorpay. Total ₹8L fits within ₹20L cap with headroom. Scored 13 (not 15) because ongoing LLM API costs scale with usage and could compress margins if usage-per-user is higher than estimated.
Team (8/10): 2 full-stack devs (WhatsApp API + backend) + 1 vernacular education content lead to tune prompts and audit answers. Achievable in 4-5 months to beta. Scored 8 because vernacular prompt quality requires an educator with domain expertise who may be hard to recruit remotely.
Trend (12/15): ChatGPT's 46M Play Store downloads in March 2026 and YC India 2026 edtech cohort are concrete, recent signals. Zuoyebang's 170M-user success in China is a direct analogue. Scored 12 (not 15) because India edtech valuations post-BYJU's have investor caution and the trend source is global, not India-specific.
Moat (9/15): Vernacular prompt libraries and accuracy reputation are defensible but copyable given 6-12 months. The real moat is accumulating subject × language × difficulty-level performance data to fine-tune a smaller, cheaper in-house model. Scored 9 — moderate moat, not structural.
Economics (12/15): 87% gross margin on base plan with 50 doubts/month assumption. Viral WhatsApp sharing keeps CAC below ₹100 for early users. Scored 12 (not 15) because LLM token costs could spike if students send photo-heavy multi-part questions, and the ₹299 price point leaves limited room to absorb spikes.
Speed (8/10): Fully API-based, no hardware, no regulatory filing. First paying user achievable in 6-7 weeks from kickoff. Scored 8 because WhatsApp Business API approval from Meta takes 1-2 weeks and is outside founder control.