DPPGenie — AI-powered adaptive daily practice engine for Indian coaching institutes

1. One-liner

AI engine that auto-generates personalized daily practice problems for each student, delivers via WhatsApp, auto-grades photo submissions, and sends weekly parent reports.

2. Trend signal — why now?

• India’s online coaching market was valued at $231.6M in 2023 and is projected to reach $971M by 2034 at 14% CAGR. The growth is driven by tier-2/tier-3 adoption and personalization demand (Allied Market Research).
• DPP (Daily Practice Problem) sheets are the backbone of Indian competitive-exam coaching (JEE, NEET, UPSC Foundation, CAT). Every serious coaching institute — from Allen and FIITJEE down to the 50-student neighborhood center — assigns daily problem sets. But small institutes create DPPs manually: the faculty member picks 10–15 questions from a book, photocopies them, hands them out. No personalization. No grading follow-up. No data.
• Existing question-paper generators (Gyanam, PaperKatta, SchoolDeck, eTutor) solve the creation problem with static question banks of 250K+ questions. But they stop at PDF generation. None of them deliver, collect submissions, grade, or adapt. The workflow after the PDF is still 100% manual.
• Students in 2026 are already receiving study material via WhatsApp — coaching institutes use WhatsApp groups as their primary distribution channel. 600M+ WhatsApp users in India, 98% open rate (Fonada — WhatsApp automation for education).
• AI can now generate high-quality NCERT-aligned MCQs with solutions and difficulty tags from chapter topics — GPT-4o and Claude handle Indian exam formats well. The AI quality gap that made this impractical 2 years ago is closed.

3. The opportunity

The typical small coaching institute (50–500 students, JEE/NEET/Foundation) DPP workflow:

1. Faculty member spends 30–60 minutes selecting 10–15 questions from a reference book or past papers. Same set for every student — no personalization by weak topic.
2. Photocopies or WhatsApp-photos the sheet to the class group. Students screenshot it.
3. Students solve on paper (if motivated). Most don’t — there’s no accountability mechanism.
4. No submission collection, no grading, no tracking of who did what.
5. Parents have zero visibility into whether their child is practicing daily. They ask the institute, get vague answers.
6. Faculty has no data on which topics students are weak in — they teach the same syllabus in the same order regardless.

The result: the most important learning activity (daily practice) has zero feedback loop. The big players (Allen Digital, PW, Unacademy) solve this with their app ecosystems and dedicated content teams. The 50–500 student institute can’t afford that — but their students compete against the same exams.

DPPGenie fills that gap: the institute plugs in their syllabus and exam type, and the system runs on autopilot — daily personalized problems, WhatsApp delivery, photo-graded submissions, parent reports. No app download. No content team. No IT staff.

4. Target market

• Primary customer: Owner/director of a small-to-mid coaching institute (50–500 students) focused on JEE, NEET, Foundation (Class 8–10), or UPSC prelims. 1–10 faculty members. Based in tier-2/tier-3 cities (Kota, Patna, Ranchi, Lucknow, Nagpur, Sikar, Jaipur). Age 35–55, usually a former faculty member who started their own center.
• Why they buy: “Allen ke bacche daily DPP solve karte hain app pe. Mere students WhatsApp pe baith ke reels dekhte hain. Agar meri institute mein bhi personalized DPP aaye toh parents impressed honge aur admissions badhenge.” The sales pitch is simple: look like Allen, for ₹2,999/month.
• Rough TAM reasoning: India has ~75,000–100,000 coaching institutes and tuition centers offering competitive exam prep (FICCI-EY estimates). The “serious buyer” — 100+ students, JEE/NEET focus, ₹50K+ monthly revenue — is ~20,000–30,000. At ₹2,999/mo, 2% penetration (400–600 customers) = ₹1.4–2.2 Cr ARR ($170K–$260K). At 5% (~1,000–1,500), ₹3.6–5.4 Cr ($430K–$650K). Path to $1M+ needs the per-student pricing upsell (see business model) which pushes blended ACV to ₹60K+.
• Why now for them: (1) Parent expectations are set by Allen/PW — they demand daily practice and progress reports. Institutes without them lose admissions. (2) WhatsApp delivery is free infrastructure — no app to build. (3) AI question generation is finally good enough for exam-quality MCQs.

5. Product sketch (MVP)

• Syllabus setup: institute selects exam type (JEE Main, NEET, Foundation Class 9–10) and current chapter coverage. System maps to NCERT + coaching-standard topics.
• Auto-generated DPPs: AI generates 10–15 MCQs per day per batch, tagged by topic, difficulty (easy/medium/hard), and NCERT chapter. Faculty can review/edit before release. Mix adapts based on aggregate batch performance.
• WhatsApp delivery: DPP lands in each student’s WhatsApp at 6 PM (configurable). Clean image format — question numbers, options, space for answers. No app needed.
• Photo submission + auto-grading: student solves on paper, photographs the answer sheet, sends back on WhatsApp. AI reads the marked answers (OMR-style or handwritten option letters) and auto-grades. Result returned in 30 seconds.
• Student performance dashboard: topic-wise accuracy heatmap, streak tracking (how many consecutive days the student submitted), time-series progress chart. Accessible to faculty on mobile web.
• Weekly parent report via WhatsApp: auto-generated card to parent — “Rahul completed 5/7 DPPs this week. Strongest topic: Kinematics (85%). Weakest: Organic Chemistry (52%). Rank in batch: 12/45.”
• Adaptive difficulty: if a student scores >80% on a topic 3 days in a row, difficulty increases. If <40%, difficulty drops and a “concept revision” micro-lesson link is suggested.

6. AI angle — what’s load-bearing

• Question generation: not pulling from a static bank (that’s what Gyanam does). AI generates novel questions aligned to NCERT and exam patterns — with distractors that test common misconceptions. This is the core value: infinite fresh content without a content team. Faculty reviews, but doesn’t create from scratch.
• Answer-sheet OCR: student photographs a handwritten answer sheet. AI reads which option they marked for each question (A/B/C/D) from a photo of messy student handwriting. Without AI, you need an OMR scanner or manual grading — neither works at scale on WhatsApp.
• Adaptive difficulty engine: adjusts the mix of easy/medium/hard questions per student based on their rolling accuracy per topic. Simple ML on performance data — not deep RL, just effective.
• Parent report generation: aggregates performance data into a clean, jargon-free WhatsApp card in Hindi or English. AI writes the narrative (“Rahul improved 15% in Physical Chemistry this week”).

Strip the AI and you have a WhatsApp group where the teacher sends a static PDF — which is exactly what they already do. The AI is what makes it personalized, graded, and reported.

7. Localization angle

India-only — the product is the localization:

• NCERT + Indian exam alignment: JEE Main, NEET, Foundation syllabus mapping. Question patterns mirror NTA style (MCQ, numerical value, assertion-reasoning). No global EdTech tool does this.
• WhatsApp as the delivery surface: Indian coaching students don’t install study apps (or install and forget). WhatsApp is where they already receive announcements, PDFs, and video links from their institute. Delivery on WhatsApp = 98% open rate.
• Hindi + English bilingual: questions and parent reports in both languages. Many tier-2/tier-3 students prefer Hindi-medium questions.
• Parent communication via WhatsApp: Indian parents actively monitor their child’s coaching progress — but currently have no visibility. The weekly WhatsApp report card is the feature that drives admissions and retention for the institute.
• ₹2,999/mo pricing: positioned as an operational expense, not a tech investment. Comparable to one month of a part-time content writer’s salary.

Allen Digital, PW, and Unacademy have their own DPP engines — but they serve their own students on their own apps. DPPGenie is for the independent institute that competes against them.

8. Business model — path to $1M–$5M ARR

• Pricing: ₹2,999/mo base (up to 100 students, 1 exam type) → ₹5,999/mo Pro (up to 300 students, multiple exams, adaptive difficulty, parent reports) → ₹9,999/mo Premium (500+ students, custom branding, API access for their own app).
• Per-student add-on: ₹99/student/month above the tier cap (scales with enrollment).
• ACV: blended ₹50,000/year ($600) when per-student pricing kicks in.
• Path to $1M ARR: 170 institutes at blended ₹50k = ₹85L ≈ $1M. That’s 0.6% of the 30K serious-buyer TAM.
• Path to $5M ARR: ~700 institutes at blended ₹60k (mix shifts to Pro/Premium + more students per institute) = ₹4.2 Cr ≈ $5M. 2.3% penetration, achievable in 24–30 months.
• Expansion path: (1) Content marketplace — top faculty upload their proprietary questions, earn royalties when other institutes use them. Platform takes 30%. (2) Student-direct tier — self-study students subscribe directly at ₹199/month (bypassing the institute). B2C at scale. (3) Test-series white-label — institutes sell branded “DPP test series” as a product to students in other cities. DPPGenie powers the backend.

9. Go-to-market wedge — first 100 customers

• Motion 1 — Kota coaching cluster blitz (customers 1–30): Kota has 200+ coaching institutes within a 5km radius, most of them 100–500 student operations competing against Allen and Resonance. The pitch: “Your competitor’s students get personalized DPPs on an app. Yours get a photocopy. Fix that for ₹2,999/month.” Visit 50 institutes in 2 weeks, demo on-the-spot, offer 1 month free. Convert 30.
• Motion 2 — Coaching-owner WhatsApp communities (customers 30–70): There are 15–20 large WhatsApp groups and Telegram channels where coaching-institute owners share tips, syllabus updates, and marketing strategies (Coaching Mantras, JEE/NEET Faculty Network, etc.). Drop a 2-minute demo video + a “free trial for first 50 institutes” offer. These communities are tight-knit and word-of-mouth compounds fast.
• Motion 3 — YouTube faculty influencer tie-ups (customers 70–100): Hindi-speaking faculty YouTubers (Physics Wallah spinoff educators, Mohit Tyagi-type channels) have 100K–2M subscribers of students and aspiring/current faculty who run their own institutes. ₹40–60K per branded integration. One viral demo = 200+ signups.

This buyer is in Kota, in coaching WhatsApp groups, and on Hindi YouTube — not on ProductHunt.

10. Build complexity — justification

Medium. MVP needs: LLM (GPT-4o/Claude) with exam-specific prompting for MCQ generation + solution generation, a curated NCERT topic-tree (manual but one-time, ~2 weeks for JEE + NEET Physics/Chemistry/Biology/Maths), WhatsApp Business API for delivery + photo collection, vision model for answer-sheet OCR (handwritten A/B/C/D recognition — simpler than full handwriting OCR), a mobile-first web dashboard for faculty + performance analytics, and a report-card image generator for parent WhatsApp messages. Two builders (one full-stack, one AI/pipeline) can ship a credible v1 for JEE Main Physics + Chemistry (2 subjects) in 10–12 weeks. Scope risk: trying to cover all subjects and all exams on day one. Ship JEE Physics first, expand weekly.

11. Gating checklist

Gate	Pass?	Note
Legal in target market	✅	No regulatory barrier. Generating practice questions is not regulated.
Ethical — no harm / dark patterns	✅	Helps students practice more effectively. No cheating facilitation — these are practice problems, not exam papers.
Market exists (evidence above)	✅	75K–100K coaching institutes, DPPs are universal practice, Allen/PW set expectations.
1–5 person team can build this	✅	2 builders, 10–12 weeks for 2-subject MVP.
Launchable with <$50K / ₹40L	✅	LLM API costs + WhatsApp API + 2 developers for 3 months ≈ ₹20–30L.

12. Feasibility score

Axis	Weight	Score	Notes
Problem intensity	20	16/20	Universal practice (every institute assigns DPPs), zero automation today. Parents demanding Allen-level personalization from small institutes. The gap between what parents expect and what the institute delivers is widening. Dock 4: the institute owner feels the pain indirectly (through lost admissions), not as a daily burning problem.
Demand evidence	15	10/15	Allen/PW DPP systems set expectations. 75K+ coaching institutes exist. Static question-bank tools (Gyanam, PaperKatta) prove partial demand. But no explicit “I want this” signal from institute owners online — this buyer doesn’t post on forums.
Build feasibility	15	11/15	LLM MCQ generation is proven but needs rigorous quality control for competitive exams. Answer-sheet OCR from handwritten photos is non-trivial (messy student handwriting, poor lighting). NCERT topic tree is manual grind. 10–12 weeks is realistic but tight.
Distribution clarity	15	12/15	Kota cluster is the single densest pool of coaching buyers on Earth — physically visit 50 in 2 weeks. WhatsApp communities of coaching owners are a real channel. But each sale is consultative (demo, convince the owner, onboard faculty), not self-serve.
Revenue mechanics	15	11/15	₹2,999 base + per-student pricing pushes ACV to ₹50K+. 170 institutes for $1M is achievable. Per-student add-on is the key lever — without it, ACV is too low. The math works if per-student pricing is adopted.
Time to first revenue	10	5/10	10–12 week build + Kota blitz + onboarding + 1-month free trial = ~5 months to first revenue. Consultative sales cycle adds time.
Defensibility	10	5/10	Proprietary question corpus + student performance data compounds over time. Faculty-contributed content marketplace locks supply side. Allen/PW won’t sell their DPP engine to competitors. Copyable by a funded EdTech but niche enough for a bootstrap.
Total	100	70/100

13. Qualitative modifiers

Founder-fit tags

technical-heavy · sales-heavy · content-heavy · domain-expertise-required

The AI question generation needs strong prompting and quality evaluation (content-heavy + technical-heavy). Selling to coaching-institute owners requires in-person demos and relationship building (sales-heavy). Understanding JEE/NEET exam patterns, difficulty calibration, and NCERT alignment requires someone who has either taught or taken these exams (domain-expertise-required).

Key assumptions to validate (3–5)

1. Assumption: AI-generated JEE/NEET MCQs are usable with minor faculty edits (>90% accuracy). How to test: Generate 100 Physics MCQs, have a current JEE faculty member grade for correctness, difficulty, and exam-alignment.
2. Assumption: Coaching-institute owners will pay ₹2,999/mo for a DPP tool. How to test: Visit 15 institutes in Kota, demo the concept, ask for verbal ₹1 commit.
3. Assumption: Students will actually photograph and submit their answer sheets via WhatsApp. How to test: Run a 1-week manual pilot with one institute — send DPPs via WhatsApp, measure submission rate.
4. Assumption: Answer-sheet OCR can reliably read A/B/C/D from student photos. How to test: Collect 50 student answer-sheet photos, run through vision model, measure accuracy.
5. Assumption: Per-student pricing is accepted by institute owners (critical for ACV). How to test: In the Kota visits, pitch both flat-rate and per-student pricing, gauge reaction.

Risk flags

1. Content quality risk: AI-generated competitive-exam questions have zero tolerance for errors. A wrong answer key or a dimensionally inconsistent Physics problem destroys faculty trust instantly. Faculty review must be mandatory in v1.
2. Student compliance risk: If students don’t submit answers via WhatsApp, the grading and adaptive loop breaks. The product becomes an expensive PDF sender. Gamification and parent pressure are the levers.
3. Price sensitivity: A 200-student institute in Patna earning ₹3L/month may resist ₹2,999/month. The ROI pitch must be concrete: “retain 5 students who would switch to Allen Online = ₹25K/month saved.”

14. Structured verdict

Score:                  70/100
Verdict:                GO
Confidence:             Medium
Best-fit builder:       Technical founder with LLM/AI experience + a JEE/NEET faculty co-founder or advisor. Must be comfortable with in-person sales in Kota/Patna.
Time to revenue:        ~5 months (10–12 week build + Kota sales blitz + trial period)
Capital to launch:      ₹20–30L ($24K–$36K)
Top 3 assumptions to validate first:
  1. AI MCQ quality >90% usable — generate 100, have faculty grade them
  2. Institute owner willingness to pay ₹2,999/mo — visit 15 in Kota, get verbal commits
  3. Student submission rate >50% — run 1-week WhatsApp DPP pilot with one institute
Kill criteria:
  - AI question quality <80% usable after faculty review = content engine isn't exam-ready
  - <6 of 15 institute owners say "bill me" = price or need doesn't justify the product
  - Student submission rate <30% in pilot = the feedback loop won't close

15. Risks & unknowns — top 3 things that could kill this

1. AI-generated questions are wrong or trivially easy. If an MCQ has two correct options, or a Physics question violates dimensional analysis, faculty loses trust instantly. Competitive-exam content has zero tolerance for errors. What has to be true: a faculty review step before every DPP release is mandatory in v1. AI generates, faculty approves. Over time, build an eval set of 10,000+ verified questions to benchmark quality. Never auto-release without review until accuracy exceeds 98%.
2. Student submission rate is too low. If students don’t photograph and send their answers back on WhatsApp, the grading and adaptive loop breaks. The product becomes a fancy PDF sender. Mitigation: gamification (streak counts, batch leaderboards), institute-enforced submission deadlines, and parent visibility (“Your child didn’t submit today’s DPP”). Social pressure from parents is the strongest compliance mechanism in Indian coaching.
3. Coaching-owner price sensitivity. A 200-student institute in Patna with ₹3L/month revenue may balk at ₹2,999/month. Mitigation: ROI pitch: “If this helps you retain 5 students who would have switched to Allen Online (₹5,000/student/month × 5 = ₹25,000/month), the tool pays for itself 8× over.” Alternatively, offer a ₹999 starter tier for institutes under 50 students and prove value before upselling.

16. Next step — 1-week validation sprint

• Day 1: Generate 100 JEE Main Physics MCQs using GPT-4o with a detailed exam-pattern prompt. Have a current JEE faculty member grade them for correctness, difficulty accuracy, and exam-alignment. Target: >90% are usable with minor edits.
• Day 2–3: Visit 15 coaching institutes in Kota (or call if remote). Two questions: “How do you create and distribute DPPs today?” and “Would you pay ₹2,999/month for a system that auto-generates, delivers via WhatsApp, auto-grades, and reports to parents?”
• Day 4: Build a Figma mockup of the WhatsApp DPP flow (delivery → student solves → photographs → graded → parent report). Record a 2-minute Hindi Loom demo.
• Day 5: Send the Loom to 10 interested institute owners from Day 2–3. Ask for verbal ₹1 commit.
• Day 6–7 — Decide: GO if AI question quality is >90% usable and ≥8 of 15 owners say “bill me” and at least 2 say “I’d pay per-student on top.” No-go if question quality <80% (AI isn’t exam-ready) or <6 verbal commits (price or need isn’t there).

Falsifiable: AI question quality <80% = the content engine isn’t ready. <6 verbal commits = institute owners don’t see this as a paying problem.