The Agentic Data Company
The Agentic Data Company builds large-scale, multilingual speech datasets that capture the full complexity of human voice—including identity, emotion, intent, hesitation, and acoustic environment—for training AI models that understand natural conversation. The company offers both research and commercial access to its corpora, with a streamable sample dataset, Open Yap 1K, available for evaluation before full access is granted.
- Artificial Intelligence
Funding
Raised to date
$500KRaised to date based on public sources. This may differ from the amount the company actually raised and is based only on what is publicly available on the internet.
Founders
Founder details are not available yet.
Product
Problem
Speech is the hardest modality for AI because it carries not only words but also identity, emotion, intent, hesitation, and the texture of a room. Most speech datasets are narrow, capturing only a fraction of how people actually talk, which limits the ability of voice models to understand natural conversation across languages and contexts.
Solution
Voice provides large-scale, multilingual speech datasets that capture the full complexity of human speech, including identity, emotion, intent, and hesitation. The company records natural conversation across the long tail of how people actually talk, spanning multiple languages and acoustic environments. These datasets are designed for training AI models that need to hear the way humans do, not just transcribe words. Voice offers both research and commercial access to its corpora, with sample audio available on open datasets like Open Yap 1K for evaluation before full access is granted.
Target Audience
Primary customers are AI research labs, commercial voice technology developers, and enterprises building speech-based products that require natural, multilingual training data.
Features
- Large-scale multilingual speech corpora capturing identity, emotion, intent, hesitation, and room acoustics
- Open Yap 1K sample dataset with streamable audio for researchers to evaluate corpus quality
- Data use agreements with SHA-256 hash records for verifiable licensing terms
- Separate privacy framework for speakers recorded in corpora, distinct from website visitor data
- Access request workflow distinguishing research from commercial use cases