A reporter’s experience with an AI avatar trained on her own work
Synthesia gave a TechCrunch reporter an interactive AI avatar trained only to discuss her article on fraud at venture-backed startups. The company, valued at $4 billion, sells enterprise tools for AI avatar videos and role-play training. The reporter describes mixed feelings about digital clones after seeing how the avatar was created and used.
Synthesia, originally UK-based, reached a $4 billion valuation and reported over $100 million in annual recurring revenue. It provides enterprise tools for AI avatar videos and interactive role-play sessions, alongside an API for combining its video and voice models with other services.
For the reporter, the company created personal and interactive avatars using photos and a two-minute voice sample, both with and without glasses. The interactive version was trained solely on her article about fraud at venture-backed startups and can answer questions about that piece. Its pipeline combines speech recognition, language reasoning, voice synthesis, and Synthesia’s video model.
This development could reshape how media and public relations interact, as journalists and audiences may encounter AI avatars that speak in a person’s likeness. Enterprise training and sales teams might use similar tools for scalable practice, while workers could face pressure to consent to digital replicas. If disclosure is unclear, trust in interviews, reporting, and online identity may be tested; clear labeling and consent practices may become more important.