The Digital Twin Situation
Ever wanted to be in two places at once? Same. Recently, I visited the New York offices of Synthesia, a startup currently valued at $4 billion, to see if their AI avatar tech could actually pull off a digital version of me.
Synthesia specializes in hyper-realistic avatars. They’ve moved beyond simple video generation into "agentic" platforms. These aren't just pre-recorded clips; they’re interactive. They use a stack of voice-to-text, LLMs, and text-to-voice models to actually listen to you and spit back a response. They even let you pick your engine—think ElevenLabs or OpenAI for the voice, while their proprietary tech handles the video animation.
The Experience
I sat in a mini studio for a quick photo session and a two-minute voice recording. The result? "Digital Dom." They created two versions: a personal one that just reads scripts and an interactive one trained specifically on one of my past articles about venture capital fraud.
When I tested it, the voice was solid (it even skipped my seasonal hoarseness), but my non-tech friends found it highkey creepy. My mom thought it was "amazing," even though she failed to trick it into revealing personal secrets. The bot stayed on-script and loyal to my reporting. It’s deterministic, meaning it won’t go off the rails—which is a good thing unless you’re looking for a rogue AI moment.
Why it matters
Real talk: could this replace journalists? Probably not. The core of my job is human connection and trust—things that don't exactly translate to a script-reading bot. However, for corporate training or handling repetitive Q&As after a vacation, cloning yourself lowkey makes sense. It’s definitely giving sci-fi, and while it’s not perfect, it’s way less jarring than a physical humanoid robot. If the vibes get weird, I can just log off. That's the beauty of being a digital avatar.
