Google Vids Gets Personal AI Avatars and Gemini Omni Video Editing

Google announced on July 16, 2026 that Google Vids, its AI-assisted video tool inside Google Workspace, can now generate a personal digital avatar from a selfie and a short voice recording. At the same time, Google is bringing its Gemini Omni multimodal model into Vids, letting users build or edit videos using written prompts alongside reference images. The changes push Vids from a workplace presentation helper into something closer to a full AI video creation platform, putting it alongside specialist tools like HeyGen and Synthesia.

What happened

Feature	Detail
Personal AI avatar	Created from a selfie and voice recording uploaded by the user
AI model added	Gemini Omni (multimodal: text prompt + reference images)
New editing capabilities	Background swap, lighting correction, effects, step-by-step edits
Safety measures	Avatar tied to Google account, invisibly watermarked with SynthID
Eligibility	Users aged 18 and over in certain regions only
Platform	Google Workspace

Google Vids started life as an AI-powered tool for building workplace presentation videos. The July 16 update significantly expands that scope. Users can now upload a selfie and a voice clip to generate a digital avatar that replicates their appearance and voice, making it possible to produce talking-head videos without sitting in front of a camera.

The Gemini Omni integration adds a second major capability. Omni is Google’s multimodal model, meaning it can process text and images together. Inside Vids, that translates to: write a prompt, attach reference images, and Omni assembles the video. It also handles background replacement, lighting corrections, and visual effects on footage recorded on a phone.

Step-by-step editing changes the workflow

Previously, AI video tools often required regenerating a clip from scratch to make any change. Gemini Omni inside Vids now supports incremental edits, so a user can adjust one element, review the result, then adjust another. For business users producing things like onboarding videos or product walkthroughs, that iterative loop reduces wasted time considerably.

Why it matters

Google is positioning Vids as a Workspace business tool, aimed at company updates and training content. But the personal avatar feature puts it in the same category as dedicated AI video platforms: HeyGen, Synthesia, Captions, and D-ID all offer avatar-based video creation and have built significant user bases doing exactly this.

The difference is distribution. Google Vids is already inside the Workspace subscription that millions of businesses use daily. A team that already pays for Google Workspace does not need to sign up for or budget for a separate avatar video service. That bundling pressure is something the standalone AI video startups will feel.

On the safety side, Google says avatars are locked to the account holder’s likeness, connected to their Google account, and embedded with SynthID (Google’s invisible AI watermarking system). Personal avatar access is also gated to users who are 18 or older and located in supported regions. Those guardrails are stricter than some competitors have applied at launch, though SynthID watermarks are invisible and not a guarantee against misuse.

For businesses already exploring AI integration across their content workflows, Vids represents a low-friction entry point. No new vendor, no separate login, no extra line item in the budget. The question is whether Google’s avatar quality matches what specialist tools deliver today.

Our take

Google has a history of shipping features into Workspace that look competitive on paper but lag behind focused products in execution. Vids will need to prove its avatars and Omni-generated video are good enough to actually use in client-facing content, not just internal demos.

That said, the step-by-step editing feature is the most practically useful addition here. Regenerating an entire video because one background element was wrong is a genuine pain point, and fixing that matters more to a real content workflow than avatar novelty.

If you manage social media content or video production for a business, it is worth testing Vids now. The avatar feature alone could replace a camera setup and recording session for routine internal communications. Just do not expect it to replace a polished brand video production anytime soon.

We covered a related shift in AI hardware earlier this month when OpenAI announced its first hardware device, another sign that the major AI labs are racing to own more of the end-user experience. Google is doing the same thing here, but through software bundled into tools people already pay for.

What to do about it

1. Check whether your Google Workspace plan includes access to Google Vids and confirm your region is supported.
2. Record a short test avatar using a neutral background selfie and a clear voice clip to see output quality before committing to a use case.
3. Try Gemini Omni’s background swap on an existing phone-recorded video to evaluate whether it replaces your current editing step.
4. Compare one Vids-generated video against your current tool (HeyGen, Synthesia, or similar) on the same script before switching workflows.

The practical first step is simply testing it against a real piece of content your team already produces.

Source: TechCrunch · AI