Video-generation startup Synthesia created the first interactive digital avatar of a journalist, allowing users to talk directly with a custom digital twin. The author visited Synthesia's new New York office in September, where the company recorded photos and a two-minute voice sample to build the interactive avatar. The digital twin was specifically trained on the author's published story regarding venture-backed startups and fraud, responding only to questions related to that research.
The creation process involved capturing imagery inside a mini film studio and utilizing a complex technology stack combining voice-to-text, agentic language, text-to-voice, and video models. Synthesia's platform integrates its proprietary video and voice models alongside third-party alternatives from labs like Cartesia, ElevenLabs, Google, and OpenAI. The company currently offers three primary product categories, including standard scripted video creation platforms, interactive agentic tools like Roleplay Sessions, and developer-focused API services.
This development highlights the expanding capabilities of enterprise digital avatar technology in professional media and communication workflows.