Google Vids can build a personalized AI avatar of you from a selfie and a voice recording, then cast that digital you as the on-screen presenter in Gemini Omni–generated videos inside Google Workspace.
Last verified · 2026-07-18 · by Moe Ameen
Google Vids is the AI video creation app inside Google Workspace, and as of July 16, 2026 it can build a personalized AI avatar of you. Instead of picking one of Google's stock presenters, you upload a selfie and a short voice recording, and Vids generates a digital version of yourself that looks and sounds like you. Type a script and that avatar delivers it on screen — no camera, no studio, no re-shoot. Google's own framing for the launch was blunt: you can now "star in" your videos.
The avatar plugs into Gemini Omni, the multimodal model Google has been folding into Vids. You cast your avatar into a generated clip and direct it with plain text — have it walk, talk, or use objects in the scene — generate scenes from a written prompt plus reference images, do image-to-video, and make iterative edits to backgrounds, lighting, and effects without regenerating the whole thing. It builds on a June 17, 2026 update that expanded Vids from 23 to 53 preset avatars across photorealistic, 3D-cartoon, and graphic-novel styles, brought language support to 24 languages, and gave avatars 30+ Gemini Audio voices.
Google gates the personal-avatar feature behind identity and safety controls. It requires a Google AI Pro, Google AI Ultra, or eligible Workspace business plan; the avatar is tied to your Google account and restricted to your own likeness; every generated clip carries an invisible SynthID watermark so it can be identified as AI; and access is limited to certain regions and users 18 or older. To create an avatar of a real person, Vids runs a likeness-capture verification step — you record your own face and voice, and the avatar is restricted to the account holder's likeness — as a consent and anti-deepfake control. Google is rolling this out in stages, so treat exact availability, regions, and edition coverage as a snapshot and confirm the current state in Google's own release notes.
The honest read: Google Vids is a strong, low-friction way to produce a clean spokesperson video of yourself inside Workspace — explainers, training clips, internal updates, a sales walkthrough. What it produces is a finished video that lands in Drive, Slides, or a share link. It is a video maker, not a distribution engine: it doesn't caption for a feed, reframe across aspect ratios, hold a persona voice across a calendar, clip a long recording into shorts, or publish to a single social platform.
Treat the Google Vids export as raw source footage, not a finished post — because that's exactly what it is. Vids hands you one polished 16:9 MP4 of you talking; the work of turning it into a scroll-stopping feed week is untouched, and that's the seam Kompozy is built to close. Drop the Vids clip into Kompozy and Clipped Shorts cuts the strongest 20–40 seconds into vertical shorts with burned-in, word-synced captions, auto-reframed to 9:16 for Reels and TikTok, 1:1 for the grid, and 16:9 for YouTube — so one avatar take becomes a handful of feed-native clips instead of a single file in Drive. From that same take Kompozy spins the transcript into a Carousel, Quote Graphics, native Text Posts, a Blog Article, and an Email Newsletter, every asset held to one voice by your Persona Brief and banned-word filters so the written pieces sound like the on-camera you.
The bigger unlock is cadence. Vids' avatar is tied to your Google account and delivers one script per render — great for a one-off, but a content calendar needs a recognizable identity you can fire again and again without re-uploading a selfie every time. Kompozy runs that as an AI Influencer persona pool (1:N, one primary): Persona Shorts for captioned talking-head shorts, Persona HeyGen for longer multi-scene avatar video, Persona VFX HeyGen with a generative hook prepended, and Persona Frames that composites the avatar as a movable layer inside a brand-exact HyperFrames template — and Gemini face-lock carries that same face into still formats like Persona Photos, Persona Infographic, and Persona Tweets. So you can prototype the look in Vids, then let Kompozy produce the recurring version at volume and, with Autopilot and a per-post review pipeline, schedule and publish it across nine social platforms plus blog and email from one queue.
Announced July 16, 2026, it lets you build a personalized AI avatar of yourself inside Google Vids from a selfie and a short voice recording — a digital version of you that looks and sounds like you. You then cast that avatar as the on-screen presenter in Gemini Omni–generated videos and direct its actions with a text prompt, so you can "star" in a video without filming it.
A selfie and a short voice recording, plus an eligible plan — Google AI Pro, Google AI Ultra, or a supported Workspace business plan. To create an avatar of a real person, Vids runs a likeness-capture verification step (recording your own face and voice) as a consent and anti-deepfake check. The avatar is tied to your Google account and your own likeness, access is limited to certain regions and users 18 or older, and every clip is SynthID-watermarked.
Both generate a talking-head avatar from a likeness and a script. Vids lives inside Google Workspace and exports a 16:9 video to Drive; HeyGen is a standalone avatar-video platform. Neither, on its own, captions for a feed, reframes to 9:16 vs 1:1, holds a persona voice across a whole calendar, or publishes to social platforms — that distribution layer is what a content engine like Kompozy adds on top of either one.
Not from Vids. It produces a finished video inside Google Workspace but does not reframe to vertical, burn in feed captions, or publish to social platforms — the export lands in Drive or a share link. To get it onto TikTok, Reels, Shorts, and the rest, run the clip through a tool that clips, reframes, captions, and schedules it. Kompozy does that and can also generate the recurring persona version at volume.
Yes. Every clip made with a personalized avatar carries an invisible SynthID watermark, Google's AI-provenance marker, and the avatar stays tied to the account holder's likeness — part of the consent and anti-deepfake controls around the feature.