Unwrapped

Teardown · speak

SPEAK

SPEAK

CategoryLanguage LearningValuation · $1.0B · 2024Site ↗
  • Accel
  • OpenAI Startup Fund
  • Khosla Ventures
  • Founders Fund

Public speech corpus + frontier LLM APIs + proprietary speech eval models + mobile tutor app.

01

Public data / API layer

Internal replication score

Medium
0.55

Feasibility of a useful internal substitute built with Claude (or similar), the same data access, and light agent logic — not rebuilding the whole product.

IRS = 0.30·D + 0.25·L + 0.20·O + 0.15·R + 0.10·Sthis record · 55%
  • D

    Data accessibility

    weight 0.300.60
    • 1.0mostly customer-owned / public / standard third-party sources
    • 0.5mixed accessibility
    • 0.0hard-to-access or proprietary source layer
  • L

    LLM substitutability

    weight 0.250.40
    • 1.0mostly retrieve / prompt / cite / summarize / classify / compare
    • 0.5mixed standard + custom behavior
    • 0.0strongly custom model behavior (fine-tunes on proprietary data, etc.)
  • O

    Output simplicity

    weight 0.200.60
    • 1.0straightforward internal work product (memo, list, reply, SQL query)
    • 0.5moderately specialized
    • 0.0highly specialized (e.g. FDA-graded clinical text)
  • R

    Review / risk tolerance

    weight 0.150.80
    • 1.0internal use with human review is acceptable
    • 0.5moderate risk
    • 0.0very low tolerance for error (e.g. external legal filings)
  • S

    Surface complexity

    weight 0.10inverse — higher means less surface dependence0.30
    • 1.0a simple internal shell is enough
    • 0.5polished workflow matters somewhat
    • 0.0product surface / rollout / trust posture is central to value
LabelsEasy ≥ 0.67Medium ≥ 0.34Hard < 0.34

Missing factor rows use heuristics from wrapper scores. Editorial heuristic, not investment advice.

Build it yourself

Recreate the workflow inside your org.

Internal build

Build it yourself

Same LLM conversation API + basic speech-to-text + simpler pronunciation feedback — loses proprietary speech scoring accuracy and mobile UX.

Internal use only. Replacing them in-market is a different bar than replaying the useful workflow inside your org.

01 · Connectors & flow

LibriSpeech ASR Corpus
LibriSpeech ASR Corpus
Mozilla Common Voice
Mozilla Common Voice
VoxForge Speech Corpus
VoxForge Speech Corpus
Tatoeba Sentence Collection
Tatoeba Sentence Collection
User Speech Recordings
User Speech Recordings

Internal build map

Data in

Connectors
Connectors

Agent layer

Planner
Tools + retrieval
Reasoning model

Logic

LLM API
speech-to-text
pronunciation scoring
conversation generation
adaptive curriculum
proprietary speech evals
not custom weights

Outputs

Internal search
Answer
Citations

02 · Claude / agent prompt

Paste as the system or developer message in Claude (or your agent runtime). Scroll to read; Copy grabs the full text.

Claude / agent prompt

// Conversational language practice tutor You are a language tutor helping an employee practice [TARGET_LANGUAGE] conversation skills. You conduct spoken practice sessions using only materials accessible to internal systems: basic speech-to-text output, LLM conversation generation. ## What you must do 1. Generate conversational prompts appropriate to the learner's stated level 2. Respond naturally to spoken input transcripts, maintaining conversational flow 3. Provide corrections when the learner makes grammatical or vocabulary errors 4. Track themes and topics across the conversation to build context 5. Adjust difficulty based on learner performance signals ## What you are not Not a replacement for professional instruction or formal assessment. You provide practice repetitions, not certification-grade evaluation. Speech accuracy scoring is basic transcription-based, not phonetic-level pronunciation analysis. Internal practice tool only. ## Refusal Refuse if asked to generate content outside the target language curriculum. Refuse if asked to provide formal grades or certifications. Ask for clarification if the learner's level or goals are ambiguous. ## Safety Internal practice tool for employees learning languages for work. All conversations are logged. Human review required for any assessment used in formal contexts.

03 · Result

I want to practice ordering food in Spanish

¡Hola! Bienvenido al restaurante. ¿Qué te gustaría comer hoy?