Here’s what happened.
Tavus’s Image-to-Replica workflow starts from a suitable human-like portrait rather than a recorded training video. The resulting character connects to its conversational video platform. The company describes uses in training, coaching and customer engagement.
The We Are So Done take
First we put our faces on the company website. Then we taught the website to answer questions. Now the face can do the answering. Excellent. We can finally skip the meeting and still look engaged.
For those of us already clicking
Give the picture a job
- Access
- Tavus developer platform; the image needs a clear, human-like face.
- Start here
- Use a portrait you have permission to use. Give the avatar one small brief, then test whether its answers stay within it.
- The small print
- A convincing face does not ensure a useful answer. The application still needs a defined role, appropriate knowledge and testing. Tavus says video remains the higher-fidelity route for reproducing a specific person.
Follow the evidence.
Source-based reporting. No independent hands-on test by this newsroom.
Sources checked 2026-09-23 · Editorial updated 2026-09-23