# Can You Build a LingoTask Competitor? — Honest Assessment

**Date:** 20 June 2026
**Context:** Roger asked whether he could build competitive software against LingoTask

---

## The Short Answer

**Not their full product. Yes, a focused competitor.**

Building what LingoTask has today — multi-subject, 99% handwriting OCR, native app + web, 190+ schools, government grant alignment — would take **2–3 years and multiple million HKD** with a dedicated ML engineering team. That's not a side hustle. That's a funded startup.

But building a **narrower, smarter** competitor — one that serves learning centres instead of schools — is very achievable, especially since you already have the distribution channel (Simply English's students) and curriculum expertise.

---

## Why Full LingoTask Is Hard to Beat

### 1. The OCR Moat

They hit **99% handwriting accuracy** for English + Simplified Chinese + Traditional Chinese. This took:
- CUHK's speech/language lab (decades of research)
- Years of real classroom data (every essay submitted trains the model)
- Multiple failed attempts — they learned 90% = useless in classrooms

**Can you replicate this?** Not easily. Off-the-shelf OCR (Google Cloud Vision, Azure, Tesseract) won't hit 99% on messy HK student handwriting. You'd need to fine-tune models with thousands of labelled HK student essays.

**Workaround:** Skip handwriting OCR for MVP. Have students type essays or teachers scan + submit. Many AI marking tools start this way.

### 2. The Rubric Alignment

Their marking aligns to DSE/TSA/HKAT/IELTS with **90% human-marker agreement**. This isn't just "grade this essay" — it's:
- Subject-specific rubrics (content, language, organisation, tone)
- Level-appropriate feedback (beginner vs advanced students)
- Exam board familiarity

Building this takes a fine-tuned system validated against actual exam scripts.

**Can you replicate this?** Partially, with modern LLMs. GPT-4o, Claude 4, and DeepSeek can do impressive DSE-aligned marking with good prompt engineering. The gap will be in **consistency** — the AI will be ~80–85% aligned vs their 90%. For your own centres, that might be good enough.

### 3. The Team

| LingoTask Has | You Have |
|---|---|
| MIT/PhD researchers from CUHK | Real classroom experience (Simply English) |
| A dedicated 10+ person eng team | ? |
| Years of R&D funded by grants | Side hustle budget |
| Full-time school sales team | Your existing network |

### 4. The Distribution Moat

They're plugged into the **EDB「智啟學教」HK$500K grant**. Schools pay effectively $0. Your competitor would need to justify a separate budget line. That's a much harder sell.

---

## The Real Opportunity — What You *Could* Build

Instead of competing head-on with LingoTask for schools, build for the **learning centre market** they're ignoring:

### Option A: Simple AI Marking Tool for Centres (MVP — 1–2 months)

| Feature | Build With |
|---|---|
| Upload student essay (typed or photo) | Simple web app |
| AI marking with DSE/IELTS rubric | Claude 4 / GPT-4o prompt engineering |
| Grammar + vocabulary suggestions | Existing LLM capabilities |
| Per-student progress tracking | Database + basic dashboard |
| Teacher review + override | One-click accept/edit |

**Cost:** ~$500–2K/month in API costs for your own students
**Build time:** 1–2 months with a good freelance developer
**Target:** Your own Simply English centres first → then sell to other learning centres

**Weakness:** No handwriting OCR (students type or teachers transcribe). But honestly, for a learning centre, this is acceptable.

### Option B: Speaking Assessment Tool (3–4 months)

LingoTask's speaking evaluation is one of their strongest features. But with modern voice AI:
- Whisper for transcription (handles HK accent decently)
- LLM-based evaluation against DSE speaking rubrics
- Practice scenarios specific to DSE Paper 4 (group discussion)

**This could be a standalone product** — HK has thousands of DSE students who need speaking practice but can't get enough live sessions.

### Option C: Writing Feedback as a Service (2–3 months)

The simplest entry point:
1. Students submit essays through a simple interface
2. AI gives rubric-based scores + actionable feedback
3. Teachers review and finalise
4. Monthly subscription for learning centres

**Pricing:** HK$1,000–3,000/month per centre (vs LingoTask's HK$29K/year for a school)
**Target:** 50–100 learning centres in HK → HK$600K–3.6M/year revenue

---

## The Smarter Strategy

Don't build LingoTask. **Build the "AI teaching assistant for learning centres"** — a lighter, cheaper, centre-focused product:

```
LingoTask:     Schools | All subjects | 99% OCR | HK$49K/yr
Your product:  Centres | English only | Typed OK | HK$1-3K/mth
```

### Why this works

1. **You have distribution** — Simply English is your own lab. Build it for yourself first, sell it when it works.

2. **Learning centres are underserved by AI** — LingoTask is priced for schools. Most HK learning centres can't afford HK$29–49K/year. There's a gap.

3. **You understand the problem** — You run the tutoring business. You know what teachers actually need.

4. **Cheap to start** — Modern LLMs make the core technology (essay evaluation) commoditised. The value is in the UI, workflow, and curriculum alignment — not the AI model itself.

5. **Scalable** — 3 centres → 10 → 50. Each one pays monthly.

### The Build vs Partner Decision

| Path | Time to Launch | Cost | Control | Risk |
|---|---|---|---|---|
| Build from scratch | 2–3 months | ~$20K HKD freelance dev | Full | Highest |
| Use existing API + wrapper | 1–2 months | ~$5K HKD | Medium | Medium |
| Partner with LingoTask (resell) | Weeks | None | None | Lowest |

If you just want to get AI marking into Simply English **fast**, partner with LingoTask or another existing tool and wrapper it.

If you want a **new revenue stream**, build your own lightweight version for learning centres.

---

## My Take

You can build a competitor — just not **their** competitor.

Target the **learning centre market**, not schools. Keep it simple: English writing feedback + speaking assessment. Use existing LLMs. Launch in 2 months. Price at HK$2K/month per centre. You won't beat LingoTask's 99% OCR or DSE alignment, but you don't need to — you just need to be good enough for tutoring centres at 1/10th the price.

**The window is opening.** AI marking tools will be table stakes in HK education within 2 years. If you enter now, even at a small scale, you build the data and reputation before the market gets crowded.

Want me to map out the technical build plan for Option A (simple AI marking tool)?