Google Vids Gemini Omni Product Demo: Generate, Edit, and Personal Avatar

To make a believable product demo in Google Vids, split the work like this: Gemini Omni generates the opening and touches up the footage, real screen recordings show the buttons, pricing, and results, and an avatar handles the voiceover. All three go into a single Vid. This guide uses a 45-second CSV (comma-separated values) import demo as an example, walking through entry checks, generation, editing, and export.

As of 2026-07-17, Gemini Omni and the personal avatar are still rolling out gradually. Rapid Release domains began on 2026-07-16; Scheduled Release domains begin on 2026-08-05. Each rollout can take up to 15 days.

Personal avatar is English-only at launch and is not available in the European Economic Area, Switzerland, or the UK. For Chinese-language product demos, switch to Vids' preset AI avatars, real voice recordings, or a voiceover tool instead. Before you start, open vids.new to check whether the entry points have appeared in your account.

Check for These Three Entry Points First

Open vids.new in desktop Chrome, Firefox, or Windows Edge. Look for Create AI videos, Edit videos, and the avatar entry in the start menu. The third entry is labeled AI avatar in the Vids getting started guide and Personal avatar in the personal avatar creation steps—what you see in your account is what you should use. Safari can only play finished videos; it cannot edit or comment.

Check Item	Conditions as of 2026-07-17	What to Do
Rollout batch	Rapid Release starting 2026-07-16, up to 15 days; Scheduled Release starting 2026-08-05, up to 15 days	Wait for your domain's batch; prepare materials after the entry appears
Personal avatar	18+, user's own likeness, English, outside EEA/Switzerland/UK	Use preset AI avatars or real voice for Chinese narration; don't create one in restricted regions
Editing non-AI video	Not supported at launch in EEA, Switzerland, UK, Texas, Illinois	Only edit AI clips from generation history, or use Vids' regular editing and trimming tools
Admin settings	Personal avatar is ON by default, admins can disable per domain; Omni has no separate admin toggle	Ask your admin to confirm Vids and personal avatar settings

Google Workspace's personal avatar announcement and Omni announcement list all the restrictions and rollout timelines above.

The launch announcement lists the following available editions. Google may adjust plans and quotas; always check the Google Vids feature availability page and your account for the actual entry points.

Official Category	Editions Listed at Launch
Business	Business Starter, Standard, Plus
Enterprise	Enterprise Starter, Standard, Plus
Education	Education Plus
Consumer	Google AI Pro, Ultra
Other Editions	Enterprise Essentials, Enterprise Essentials Plus, Nonprofits
Education Add-ons	Google AI Pro for Education, Teaching and Learning
Other Add-ons	AI Expanded Access

For personal accounts using Google AI Pro or Ultra and adding the avatar to Ingredients, the generation help doc states this usage is currently US-only. Organization accounts need to verify edition, region, rollout domain, and admin settings together.

Break the 45-Second Demo into Four Vids Scenes

This example assumes a CSV file import demo. Build four scenes along the timeline first, then fill in the generated clips, real screen recordings, and voiceover. Product facts must use real footage; generated visuals are for atmosphere only.

Time	Scene	Material	Acceptance Criteria
0–8s	Office scene opening	Omni-generated clip	No readable product UI, customer names, pricing, or claims
8–28s	Show CSV file import	Vids real screen recording	Buttons, fields, success and failure results match the current product
28–40s	Explain limitations	English personal avatar, Chinese preset AI avatar, or real voiceover	Script approved by product owner
40–45s	Next steps	Real text cards and brand assets	Links, dates, and calls to action are verifiable

After opening a blank Vid, create four scenes and note their estimated durations. Replace the pages, buttons, and results in this example with verified real product content. Use test accounts and fictitious data for screen recordings; don't include real customer information in recordings, prompts, or feedback forms.

Generate a 16:9 Opening Clip in Create AI Videos

As of 2026-07-17, Google's English generation help doc uses the Create AI videos → Create path. If your account still shows Veo 3.1 or Create from scratch, the UI hasn't switched to the version described below—follow your account's current flow and don't mix old labels with the new Omni steps.

Select Create AI videos from the start menu, then enter Create from the sidebar. If you already have a Vid open, select AI video from the right panel.
For this horizontal product demo, choose Landscape (16:9). For vertical output, choose Portrait (9:16). To add reference images, click Ingredients (up to 7); to use an avatar and its voice, click Avatar.
Enter your prompt and click Generate. After generation, check the people, text, brand elements, and background. Adjust the prompt and regenerate if needed.
Click Insert to place the approved clip into the first scene. Closing the tab clears the current session's generation history, so insert any version you want to keep before closing.

English prompt example:

Create an 8-second opening clip for a business software product demo.
Show an operations manager reviewing an import task in a quiet office.
Use the uploaded brand-color board only as a visual reference.
Do not show readable product interfaces, customer data, prices, or legal claims.
Use a static camera, natural light, and restrained motion.

Vids currently outputs generated clips at 720p, 24fps, supporting landscape 16:9 or portrait 9:16. Google's generation documentation states that most users can generate up to 50 videos per month; some plans have different limits or shared quotas. Check the generation and quota documentation before starting bulk attempts.

Edit Only One Variable at a Time in Edit Videos

Edit videos works on generated clips or phone footage. Change only one variable in the first pass—background, lighting, style, or audio—and keep the original as a reference.

Select Edit videos from the start menu. If you already have a Vid open, you can also select AI video from the right panel, then enter Edit.
Upload a clip or select one from the current Vid. If the uploaded video is longer than 10 seconds, Omni only edits the last 10 seconds. If the target scene isn't at the end, trim it first using the regular editing tool.
Enter a single verifiable edit instruction—for example, fix only the lighting without changing the subject or composition. After generation, compare the subject's movement, on-screen text, brand marks, and audio side by side. Discard the version if product facts were altered.
Insert the approved version into the scene, and keep the original until the publishing review is complete.

Keep the framing and the person's movement unchanged.
Reduce the overexposure on the left side of the frame.
Do not add text, logos, products, or new objects.

As of 2026-07-17, Google's generation help doc states that users in the European Economic Area, UK, Switzerland, Illinois, and Texas cannot upload and edit non-AI videos; they can only edit AI clips from their generation history. If your account has generation capabilities but no Edit videos entry, check the region table and rollout batch above—don't repeatedly refresh, change prompts, or re-upload footage.

Generate English Voiceover with Personal Avatar

A personal avatar can only use the current Google Account owner's likeness. The personal avatar help doc lists the launch conditions: 18+, capture via phone or tablet, English, and an eligible region. Workspace admins can disable this feature.

Select Personal avatar from the Google Vids start menu on desktop.
Scan the QR code with your phone or tablet, and follow the on-screen instructions to record your face and voice. The account owner must complete the recording themselves, with no other faces visible in the background.
Return to your computer and click Next. Once the avatar appears in the generation panel, enter an English action description and an English voiceover script, then click Generate.
Confirm that the on-screen avatar is you, that the pronunciation and lip-sync are acceptable, and that the script content hasn't drifted. Then click Insert. To re-record or delete the avatar, go to the management menu from the avatar's more options; deletion syncs to your Google Account and affects any Gemini App already authorized to use the avatar.

The placeholder script below does not represent any real product interface. Replace the bracketed content and verify every page name, button, and result before use:

Hi, this video shows how to import a CSV customer list into [product name].
Open [customer page] and select [import button].
Choose the CSV file, then review [field-mapping step] before you continue.
After the import, check [success result] and [failure result].
Download [error report] before you correct and retry failed rows.

At launch, the personal avatar cannot read Chinese directly. Vids' preset AI avatars already support Chinese—follow the AI avatar language documentation to generate them. If you need to keep your own voice, use real recordings or the Gemini 3.1 Flash TTS text-to-speech workflow.

Combine Real Screen Recordings and Generated Clips in One Vid

Real screen recordings, real screenshots, and approved text cards are responsible for presenting product facts. The Omni opening, avatar voiceover, and captions explain the operation—they must not show model-hallucinated buttons, pricing, customer data, or success results.

To record product footage, select Record from the start menu, then choose screen-only or screen + camera. After recording, place the result in the second scene.

Layer	How to Do It in Vids	What the Owner Must Confirm
Generated footage	Insert the Omni clip into the first scene	The clip carries no product facts; no random text or unauthorized assets
Product operation	Record the screen in Vids, demonstrating the CSV import in real order	The entry points, fields, and results actually exist in the current version
Voiceover and captions	Add the approved script as avatar, voiceover, and captions	Pronunciation, dynamic facts, limitations, and call to action are consistent
Final video	Adjust scene duration, captions, and transitions before exporting	Accessibility, branding, disclosures, and archival checks all pass

After placing all materials on the same timeline, play through from the start once. Even with the sound off, viewers should be able to tell real product UI from atmospheric shots. If the two are easily confused, delete the generated product interface and use real recordings instead.

Troubleshoot Missing Entry Points or Unexpected Results

Symptom	What to Check First	What to Do
Start menu missing new entries	Rollout domain, plan, admin settings, logged-in account	Wait for the batch or ask the admin; don't create duplicate projects
Only seeing the Veo 3.1 panel	Whether Omni has rolled out to this account	Follow your account's current generation flow; don't treat the old panel as the Omni edit entry
Generated clip doesn't match reference image	Number of reference images, aspect ratio, whether the prompt asks for too many changes at once	Keep one subject and one main action, regenerate, then insert
Personal avatar recording fails	Eligibility and capture environment	Re-record per the eligibility and recording requirements; switch to a preset avatar or real recording if conditions aren't met

Google's generation help doc states that feedback may be read by humans. Do not submit account credentials, customer data, confidential business information, or contract content in feedback.

Pre-Publishing Checklist and Export

Evidence	Pass Criteria
Feature and identity	Current login account and entry points verified; personal avatar created by the account owner; account, region, language, and admin conditions documented
Product facts	Product UI, dynamic facts, limitations, and results are verifiable in the current product or official documentation
Assets and presentation	Images, fonts, music, and brand assets have usage rights; captions, pacing, contrast, and pronunciation are checked
Disclosure and archiving	Keep prompts, asset sources, review records, and the final export; add AI disclosure labels per channel and business requirements

For team review, share the Vids file using the Share button in the top right. After approval, open Quick sharing options, select Export to Drive to generate an MP4, and play the exported file through once. See the Google Vids getting started guide for the full workflow.

Google states that every AI-generated clip carries an invisible SynthID digital watermark. This watermark does not replace explicit disclosures required by distribution channels, nor does it replace internal approval for high-risk or public-facing content, such as ads, education, recruitment, and regulated industry content. Source: Google official announcement, 2026-07-16.

FAQ

I meet the plan requirements—why can't I see Omni yet?

First check whether your Workspace domain is on Rapid or Scheduled Release. Rapid Release domains began rolling out on 2026-07-16; Scheduled Release domains begin on 2026-08-05. Both can take up to 15 days. Also confirm your logged-in account, edition, and admin settings. Source: Google Workspace Updates.

Can the personal avatar read Chinese scripts?

No, as of 2026-07-17. Personal avatar is English-only at launch. For Chinese, use a preset AI avatar that supports Chinese, real voice recordings, or a voiceover tool. Source: Personal avatar help doc.

Can I create a personal avatar for a colleague or client?

No. Personal avatars are tied to a Google Account and limited to the account owner's likeness. Shared team accounts should not be used to host another person's avatar assets. Source: Google Workspace Updates.

How do Google Vids and Google Flow compare?

If your focus is product screen recordings, Workspace collaboration, scene arrangement, and internal business videos, use Vids. If you need an asset library, storyboards, complex generation controls, or repeated creative polish, check out the Google Flow tutorial. As of 2026-07-17, both Google Flow's official page and Vids list Gemini Omni support.