One-Page Brief: Gamified Language Practice (City Map + Voice Companion)
Verdict
As described, the idea is a platform for everyone to learn English, Hindi or Bengali for official or casual use:
- It starts from the learner's current level.
- It is played as scenes on a city map (park, hotel, shop, a stranger asking directions, an office), by typing, choosing or talking to a voice AI.
- Conversation lessons are saved and reused to cut AI cost.
The odds of that version reaching $10k MRR in 24 months are about 2%, and it isn't fundable in its current form. Most of the core mechanic already exists for free:
- Duolingo has placement tests.
- It ships free, scene-based Speaking Adventures.
- It is moving AI Video Call into its cheaper Super tier, at a cost it says is under $0.01 a call.
Narrowed to work-focused spoken English for Hindi- and Bengali-medium learners, the picture improves. That means a speaking placement, a map of work situations, explanations in the learner's own language, and a measurable weekly speaking score. With the 3 conditions met, the odds at month 24 are:
| MRR | ~15 hrs/week | Full-time |
|---|---|---|
| $1k | 35% | 50% |
| $3k | 22% | 33% |
| $5k | 15% | 24% |
| $10k | 7% | 13% |
| $20k | 3% | 6% |
| $50k | 1% | 2% |
For comparison, 17.3% of all new subscription apps reach $1k a month within 2 years, and 4.6% reach $10k (RevenueCat 2026).
The five things that matter
- Indians pay for spoken English, but not much per person. SpeakX reports about 200k paying subscribers at ₹299 a month, about $7.5M ARR, and profitability (Oct 2025). Stimuler charges ₹199 a month or ₹999 a year. The demand is proven. The price band is ₹199–299, so volume and low cost per learner matter more than features.
- "It starts from zero" is not true of the leaders. "It doesn't test speaking" is. Duolingo places learners with a reading and listening test, and its 148 AI-built courses are mostly A1–A2. The unmet need is the learner who understands English but freezes when they have to speak at work. The product should place and teach by speaking.
- The city map is UI, not a moat. Duolingo Adventures and Speaking Adventures already put learners in scenes with characters. Every scene-based language game found is a small indie business. The map earns its place only if its places are the learner's real, high-stakes situations: the interview, the customer call, the hotel desk.
- Your caching idea is half right. Scene scripts, branches, hints, exercises and NPC audio can be generated once and reused for every learner. Live voice replies can't, because every learner says something different. With "scripted spine, live edges", a voice minute costs roughly ₹0.5–1. Without it, a daily user on a ₹199 plan can cost more than they pay.
- Bengali is the gap, but a smaller one. Duolingo teaches English from Bengali but has no Bengali course to learn. SpeakX is 70% Hindi-speaking and has only announced Bengali. Bengali-medium English learners are less contested. Learning Bengali or Hindi as a new language has almost no measured paid demand. The closest signal is migrants in Bengaluru paying for Kannada classes.
Feature sort (headline)
- Table stakes: placement test, typing and multiple-choice exercises, streaks, a map or path of lessons, casual everyday scenes.
- Differentiators: placement by speaking, work-situation scenes, explanations in Hindi or Bengali, weekly speaking score with before/after recordings, institute seat licences with a trainer dashboard.
- Distractions: three target languages at launch, a real game world (3D, avatars, open map), a paid casual track, native iOS first.
- Liabilities: stored voice recordings or voiceprints, under-18 users, a "friend" companion persona, auto-converting trials, outcome guarantees.
Most likely cause of death
"Duolingo with a map." Learners enjoy the scenes for a week, can't see their speaking improve, and go back to the free app they already have.
Next 90 days
Do: pre-sell 50 people a 30-day WhatsApp version (daily work-scene prompt → learner voice note → AI and trainer feedback in Hindi or Bengali) at ₹149–199. Then ship a thin web app with 5 work scenes at 3 levels and a speaking score, and measure month-2 renewal.
Stop: designing the city, the avatars and the Hindi and Bengali target-language courses.