A Digital Version of Me. First, Let’s See How Julia McCoy Did It
My own experiment is still ahead of me. I am starting with First Movers and a question: can a digital voice and face also preserve how I explain things?
- Goal
- Explore a digital avatar example and plan a small, comparable test of my own material.
- Tools
First Movers · the creator’s example; HeyGen and ElevenLabs · tools in her account
- Process
- Creator’s example → one script → plan two versions → comparison criteria.
- Result
- A short introduction with Julia McCoy’s video and a test-planning prompt; my own experiment is still ahead.
I wanted to run my own experiment with creating a digital version of myself, but I have not got around to it yet. It remains on my list of things to explore. To start with, here is a concrete example: Julia McCoy from First Movers has shown how she creates a digital voice and avatar for her videos.
In the setup she describes, HeyGen provides the avatar — an on-screen representation of her appearance — and ElevenLabs reproduces her voice. A prepared script can then become a video presented by that character. Julia also describes the human work involved in writing and editing. This is her production process, rather than the result of my testing those tools. Her account on First Movers.
I already have a small point of reference in my own conversations: while refining a daily AI briefing, I asked for short items instead of long descriptions. I also wanted to know why each item mattered in practice. I would bring that preference to an avatar experiment: I would want the material to preserve how I explain things, as well as a similar voice and face.
Instead of rewriting the whole tutorial, here is the creator's video, in English:
While watching, pay attention to three things: how natural the voice sounds, whether lip movements match the speech, and the human work needed to prepare the final material. That is a better starting point for your own test than the first impression that “it looks like a real person.”
My first test would look like this: I would write a 30–60-second script, record it myself, and prepare a synthetic version using exactly the same text. Both versions would contain one concrete point and a short explanation of its use. Then I would compare clarity, naturalness, and the time spent preparing and correcting each version. I would tell the audience plainly that they were watching an avatar.
You can use the same acceptance criterion: after watching both versions, a viewer should be able to say what they learned. Resemblance to the author is just one element of a useful video.
For now, treat this video as a glimpse of what is possible. I will return to my own results when I have something to show.
Prepare a small test
If you want to turn this inspiration into a plan for your own test, you could start like this:
Help me plan a test using my own voice and avatar. Topic: [one thing I want to explain]. Audience: [who it is for]. Prepare a 30–60-second script with one concrete point and its application. I will record the same text myself and use it for the synthetic version. Suggest a simple way to compare clarity, naturalness, and the time needed for preparation and corrections. Do not assume the outcome before testing. Include clear disclosure of synthetic voice and imagery to the audience. At this stage, prepare only the plan and script, without uploading recordings to services or creating an avatar.