6. Fundability
6.1 Would I invest?
Pre-seed: No, as described. It's a gamified AI language app in a year when:
- Duolingo (58.7M DAU) is moving AI conversation into cheaper tiers and ships scene-based speaking free;
- Speak is valued at $1B;
- Loora raised $22M in September 2026;
- SpeakX raised $16M to expand into Bengali and other regional languages.
There's no data, distribution or cost edge yet.
Seed: No. There's no traction.
What would get a conversation: proof that learners measurably improve and keep paying. For example:
- month-3 payer retention of 35%+;
- a median speaking-score gain of 15%+ in 60 days;
- 5+ institutes paying for seats.
This business doesn't need investors to reach any level up to about $5–8k MRR.
6.2 What a skeptical investor says in the first five minutes
- "Duolingo has placement tests, scenes and free Speaking Adventures. Why you?"
- "SpeakX has 200k payers and just raised to add Bengali. What happens when they ship it?"
- "What's your month-2 renewal? Education has the worst annual renewal of any app category."
- "At ₹199 a month, what does a voice minute cost you, and what does a daily user cost?"
- "How do you acquire users when Seekho spends almost its whole revenue on ads?"
- "How do you know learners improve? What's the rubric, and who designed it?"
- "How accurate is your Bengali speech recognition on real learners?"
- "Why a city map? What does it do for retention that a list of lessons doesn't?"
- "Do you store voice recordings? Any under-18 users?"
- "Why this team? Has anyone here taught spoken English?"
6.3 Three proof points that would flip my answer
| # | Proof point | Threshold | Why |
|---|---|---|---|
| 1 | Learners measurably improve | Median speaking score up ≥15% in 60 days on re-placement, confirmed by a trainer on a sample | Shows it's a learning product, not a toy. It answers "same as Duolingo". |
| 2 | Payers stay | ≥35% of monthly payers still paying at month 3; blended churn ≤10%/mo | Beats the Education renewal objection. |
| 3 | Cheap distribution | ≥25% of new users from shares or referrals, and ≥5 institutes paying for seats | Answers the ad-spend dominance in the category. |
A strong fourth: voice and model cost below 15% of revenue at scale. That proves the "scripted spine, live edges" design works.
6.4 Bootstrap or raise?
Bootstrap. Costs are voice and model usage, hosting, payment fees, and a trainer advisor paid by the hour. There's no inventory, no live tutors and no sales team at the start. It is profitable per payer from day one if voice minutes are capped and scenes are cached. It runs asynchronously: learners practise on their own schedule, and only the institute line needs meetings.
Raise only if all three proof points hold and you decide to buy growth with paid acquisition. The $20k–50k levels mostly require that. Keep it as a later option, not the plan.
The versions not to fund at any stage:
- a general "learn any language" game (Duolingo's ground);
- live tutoring for children (sales-heavy, and a different business);
- an ad-funded "edutainment" subscription (the Seekho model needs ad spend close to revenue).