Google added prompt-based control of AI avatars to Vids and video export to YouTube

Google has expanded Vids — its AI video editor within Workspace — and added avatar management through text prompts. Now users can not just select a digital host, but literally describe what they should do on camera, what to interact with, and how they should look.

What's New in Vids

The main update is directable avatars. In Vids, you can now use plain language to set up a scene where an avatar speaks, moves, and interacts with objects. Google provides practical scenarios: showcasing a new product, explaining safety rules next to equipment, or recording an internal company address with an office backdrop. At the same time, the company expanded the entire workflow around such videos, so users can not only generate a character but also assemble the final video faster. As a result, the list of new features looks like this:

• controlling avatar actions through text prompt
• customizing appearance, clothing, and character background
• uploading reference images for locations and objects
• generating eight-second video clips based on Veo 3.1
• exporting the finished video directly to YouTube

A Chrome extension for screen recording with audio or video has also appeared separately. This is an important detail: Google is not trying to make Vids just a generator of talking-head videos, but is assembling a complete toolkit for corporate video production — from screencasts and AI inserts to publishing. For teams that create training, demos, or internal updates, this toolkit covers several steps in one environment, without switching between different services and with faster approval.

How It Works

The new mode allows you to set not just dialogue but also avatar behavior. Users can describe an action like "show the product in your hands," "stand next to the equipment," or "explain the steps with an office backdrop," and Vids will assemble the scene with the same speaker. For customization, you can change clothing, appearance, and background, as well as upload up to two additional images to more precisely define the location or object.

Google specifically emphasizes that the character's face and voice should remain stable when changing scenes. Essentially, Vids is moving from the category of templated videos toward a tool that covers typical business tasks without a film crew. This is evident from the examples Google itself highlights: employee training, product demos, internal announcements, and sales.

At launch, the feature is available for Workspace accounts set to English, so the international rollout will likely be gradual. But the direction is already clear: less manual editing, more video assembled from text, avatar, and several source materials.

A Bet on Workspace

Google first introduced Vids in 2024 as a service for creating work videos within the Workspace ecosystem. Since then, the product has been building features in layers: first AI avatars appeared, then 2D and 3D cartoon characters and new voice languages, and in March Google added Lyria 3 and Lyria 3 Pro music models for soundtracks and sound effects. Now Veo 3.

1 and more controllable avatars have been added to this stack, so Vids is increasingly becoming not an editor with AI features, but an AI-first editor. It's also important how Google packages access. The company gives all users 10 free Veo generations per month, and for Google AI Ultra and Workspace AI Ultra, the limit goes up to 1,000 videos.

The AI clips themselves within the editor are still short — up to eight seconds — but for presentations, instructions, and inserts, this is often enough. Plus, direct export to YouTube removes an extra step: the video can be sent to the platform in private mode, reviewed, and only then opened for colleagues or audience. Google is not the first to enter the AI avatar market: Synthesia, HeyGen, D-ID, and other services have been operating here for a long time.

But Google's approach is different — not a separate narrow product, but a built-in module within a package where Docs, Slides, Drive, and corporate collaboration scenarios already live. If this model works, Vids could take a strong position in the segment of quick work videos for onboarding, training, sales, and regular internal updates.

What This Means

Google is betting not on wow-factor, but on mass production of business videos. Prompt-controlled avatars, Veo 3.1, screen recording, and YouTube export turn Vids into a more complete tool for teams that need to produce lots of clear video content without a studio, presenter, or lengthy editing. If the product maintains simplicity and generation quality, it could become for corporate video what Docs or Slides are for text and presentations — a standard tool.