I built an interactive digital twin of myself — and you can chat with it
A TechCrunch reporter became the first person outside Synthesia to receive interactive AI avatars of herself. Built from photos and a two-minute voice recording, the avatars combine voice-to-text, an agentic language model, text-to-voice, and Synthesia's video model. Trained only on one story, they deflect off-topic questions. While Synthesia's tech is enterprise-ready, the experiment highlights an unresolved question: will audiences ever trust AI-presented journalism?
Key Takeaways
- Synthesia, a $4 billion digital avatar startup with $100M+ ARR, created its first-ever digital avatar clone of a journalist — a TechCrunch reporter — marking the first time anyone outside the company received one.
- The cloning process was simple: a mini film studio setup, multiple photos, and a two-minute voice recording produced four avatars (personal and interactive versions) within days.
- The interactive avatars combine voice-to-text, an agentic language model, text-to-voice, and Synthesia's video model; they're deterministic, trained only on the reporter's specific story, deflecting any off-topic questions.
- Reactions from friends and family were a mix of fascination and unease, highlighting the "creepy" side of digital twins of real people.
- Despite the technology being enterprise-ready, the reporter argues that journalism depends on trust that can't be outsourced to avatars, and public acceptance of AI-presented media remains an open question.
For the first time, Synthesia has built a digital avatar for a journalist — and for anyone outside the company at all. A TechCrunch reporter spent weeks with her own AI clones and came away with an intimate look at how the $4 billion avatar startup's technology actually works, and what it might mean for trust in media.
The Backdrop: A Rising Star in Digital Avatars
Synthesia, originally a U.K.-based company, sits among the hottest players in digital avatars alongside D-ID, HeyGen, and Colossyan. It crossed $100 million in ARR last year and reached a $4 billion valuation earlier this year.
The company serves enterprises with AI-avatar training videos and recently launched Roleplay Sessions, letting employees practice sales pitches with an interactive avatar that responds and scores their answers. The reporter's first encounter with this technology came when Alexandru Voica, Synthesia's head of corporate affairs, shared an interactive avatar of himself trained to answer press questions about the company.
Building a Digital Twin
At Synthesia's new New York office opening in September, the reporter agreed to be cloned. The process: a mini film studio inside the office, numerous photos, and a two-minute voice recording, all with consent. The team delivered four avatars in a couple of days — two personal ones (script-reading only) and two interactive ones, each with and without glasses.
The interactive avatar runs on a stack combining voice-to-text, an agentic language model, text-to-voice, and Synthesia's own video model to animate the face. Customers can also swap in voice models from Cartesia, ElevenLabs, Google, or OpenAI, and host avatars on any cloud or through Synthesia.
The interactive version is deterministic: trained only on the reporter's story about venture-backed startups committing more fraud, it deflects every off-topic question — even from her parents, who tried asking things "only they would know."
The Trust Question
Friends found the avatars interesting and creepy; her mother called it "amazing" before joking, "I don't remember giving birth to two of you." But the deeper question lingers: would audiences accept news presented by an avatar? One investor said no immediately, citing backlash against AI slop. The reporter's own take: journalism runs on trust, and that can't be outsourced.
She suspects her generation, Gen Z, won't get used to digital twins — everything once confined to sci-fi movies is now here. Still, she finds avatars less jarring than humanoids: if things get weird, you can always log off.
Where this stands: Synthesia's interactive avatar technology is live and enterprise-ready, but public acceptance of AI-presented media remains an open question worth watching as corporate adoption grows.
Meta description: Synthesia built its first-ever AI avatar for a journalist, trained on a single story. Inside the tech stack, the process, and what it means for trust in media.
Tags: Synthesia, AI avatars, digital twins, enterprise AI, TechCrunch
Featured image: Abstract technology style — glowing wireframe human silhouette against dark blue gradient, no people or logos.
Comments
No comments yet. Be the first to comment.