So, on the good side, the AI voices are good - BUT, you cannot fine-tune the text to speech, so a lot of words sound unrealistic.

Avatars look fairly good initially, but as others have said, once the TTS is added, the lip movements and facial jitters make them look like they have psychological disorders! Again, unrealistic.

If the video is over a few seconds, the AI creates multiple scenes and, even if you only selected a single avatar, you need to edit each and every scene individually to make changes to the avatar, such as background removal. On my plan, I have up to 30 min video, but it would take months to edit such a video as a result of this. Really poorly thought out!

Plan to refund! I think product managers need to spend time more time working on products here before releasing them. You end up giving yourself a poor reputation for allowing poor quality out at too early of a stage.