Talko: Learn English with AI

Product rot Commodity APIs

Soly Limited · Education · Free · v1.2.1

Review-rating drop: 0.41
Peak quarter: 5.00
Last 90 days: 4.59
Store rating: 4.69
Store gap: 0.10
Est. MRR: $114k

Store ratings include star-only taps from in-app prompts, which skew high. The peak and recent figures come only from people who wrote a review, which skews low. Compare each against itself: the drop is the signal, and a big gap between the two usually means heavy rating prompting is masking unhappy users.

Average rating by quarter from 43 text reviews

Faint bars have fewer than 8 reviews. Text reviews skew negative versus star-only ratings; read the direction, not the level.

Show as table

Quarter	Avg stars	Reviews
2026-Q2	5.00	4
2026-Q3	4.47	30
2026-Q4	5.00	9

Recent 1–2 star reviews

Trabajando

★ · v1.1.7 · Sep 22, 2026

Día

La publicidad es diferente al curso

★ · v1.1.6 · Sep 11, 2026

La publicidad tiene una profesora y cuando lo compras aparece otro formato totalmente diferente sin profesora sin escritura de pronunciación

No sirve

★ · v1.1.1 · Aug 6, 2026

Hablo y no me detecta nada y ni un ejercicio logré hacer. Cheque mi micrófono estaba activado.

Why it fell

Product rot

confidence 45%

There are only a few low-star reviews, and the drop from 5.0 to 4.59 is mild. Complaints point to speech detection not working and to the product not matching its ads (no AI teacher, no pronunciation transcription). These are product and quality issues, but the evidence is thin.

This is the kind worth building against: the product got worse and the feature is replaceable.

Data it depends on

Commodity APIs

Needs third-party data or services, but ones any developer can buy cheaply. Needs LLM conversation and speech-to-text/text-to-speech APIs (e.g., OpenAI, Whisper, ElevenLabs, Apple Speech).

Fixable complaints

Speech recognition fails to detect user's voice even with the microphone permission granted
Marketing shows an AI teacher avatar and pronunciation transcription that the purchased product lacks
Exercises cannot be started or completed when voice input fails, so a text or typing fallback is needed

The opportunity

What would it take to build a viable competitor, what would it cost to run every month, and a prompt to get started. Generated from the reviews on file; one LLM call per brief. Set ANTHROPIC_API_KEY for a real brief; the stub only fills in a template.