Author: Nadali. Prepared with AI assistance.
What changed
On 2 November, Google announced image uploads for custom AI avatars in Vids — virtual video presenters.
Who benefits: small teams making internal explainers or training videos. The practical takeaway is to review one short video first, rather than replace every instruction at once.
Try it without overpromising
Vids help describes AI avatar → Design a custom avatar → Upload image. Users then choose a listed voice and add a script. The avatar stays in its project; help recommends copying the project to carry it into another file.
Our suggested review process:
- Start with an approved instruction without customer data. Define one action that a colleague should understand after watching.
- Use an image whose rights have been checked. Do not present a synthetic speaker as a real employee or customer. Google’s rules prohibit deceptive impersonation and rights violations.
- Listen to the entire result: names, numbers, action order and pronunciation. Compare the narration with the approved script, rather than judging only the presenter’s appearance.
- Clearly label the AI presenter and provide a text version of the instruction. Assign someone to approve the video and maintain it.
This is a proposed workflow, not a test we conducted or a claim of measured time savings.
Access and limits
Rapid Release: gradual rollout from 30 October, taking up to 15 days. Scheduled Release: planned start on 20 November. Do not assume every account already has the button.
Availability help lists Business Starter, Standard and Plus as eligible plans. Administrators can disable Vids or generative features; allowances depend on the plan. Check the work account and its limits before preparing a series of videos.
Sources read on 2 November. The separate announcement page did not open; its full text was read in the official feed. Nadali did not test generation or access in a customer account. The cover is a reused AI illustration of reviewing materials, not the Vids interface or an output from this feature.
