Disney Replaces the AI Puppeteer: Anzellans Animatronic Aliens React to Guests on Their Own
Disney has built three Anzellans animatronic aliens from Star Wars that respond to guests without a puppeteer behind the scenes. Hidden cameras and microphones stream video and speech to Amazon AWS cloud, where a fast AI model catches a guest's gaze while larger models plan the conversation. Animators write the lines and movements in advance, and the AI only selects the right ones. For brands, this is a way to bring characters to life without the risk of stray words.
AI-processed from The Verge; edited by Hamidun News
Animatronics in Disney parks look more realistic every year, but they almost always work under one of two setups: either they repeat a memorized program, or a live operator works behind the scenes. Walt Disney Imagineering R&D, the company's research division, has shown a third option. These are three small alien robots, the Anzellans from the Star Wars universe. They notice guests and respond to them on their own, and their "brain" is not inside them but in the Amazon AWS cloud, meaning on servers Amazon rents out. For anyone building robots, voice assistants, and interactive products, this is a clear example of how to integrate AI into a physical character without losing control over what it says.
Who the Anzellans are and where you can see them
The new animatronics are called Vilo Vele, Kado Bre'a, and Jibu Bre'a. They are based on the Anzellans alien characters from a Star Wars film released this year. Each has its own temperament and quirks, so the three robots react to the same thing in different ways.
The first public showing will take place in Las Vegas at the end of this month, where event guests can talk with the characters in person. Disney acknowledges that the characters and the technology could eventually reach park attractions. The article from The Verge gives no specific dates or venues.
How it works without a puppeteer
The recent Olaf robot, which The Verge mentions, is brought to life by a person: an operator can control it even from a portable console such as the Steam Deck. The Anzellans have no operator. Each robot hides cameras and microphones that capture guests' movements and speech. The data goes to AWS, where several AI models process it at once. The work is divided between them by speed:
- a "fast" model makes instant decisions, such as looking a guest in the eye at the right moment so the character seems interested in them;
- larger models plan longer scenes and decide how the conversation should continue.
This is a familiar engineering pattern: a lightweight model handles reactions without delay, and a heavy model handles meaning. If the robot looks at a person a few seconds after they appear, the illusion of a living being breaks down, which is why the fast layer is critical here.
AI chooses, but does not compose
The most important detail: the Anzellans do not invent speech and movements from scratch. Every line and every movement that the models select and trigger was written in advance by Imagineers, Disney's engineers and animators. This keeps the characters true to their personas, including each one's individual traits. As The Verge reports, AI helps guide behavior, but the Anzellans never deviate from the script.
For a brand of Disney's scale, this is a sensible compromise. Free-form generated dialogue could produce an unexpected or inappropriate reply, especially when children are nearby. A pre-approved library of actions removes that risk, and the AI turns that library from a linear recording into a flexible reaction to a specific person.
What this means for practitioners
First, the pattern of "the model chooses from a prepared set" is useful beyond theme parks. It can underpin support systems, game NPCs, and voice interfaces where predictable behavior matters.
Second, real-time cloud processing of video and audio appears to be fast enough for interactive installations. However, the source provides no data on latency or cost.
Third, open questions remain. It is unknown how the system will behave in a dense crowd, what happens if the cloud connection is lost, and how often the model selects the wrong line. Answers will come when the characters are shown to guests in Las Vegas.
- Nov 21, 2026AG-UI 1.0 gives AI agents a common language for talking to apps: 31 event types
- Nov 21, 2026OpenAI launches ChatGPT Wallet: agents buy without ever seeing your card number
- Nov 21, 2026US Advertisers Spent $70.7 Million on ChatGPT Ads in October, Up 127% From April
- Nov 21, 2026ONNX Runtime vanishes from GitHub: Microsoft hides repositories after package hack
Key Questions Answered
Who the Anzellans are and where you can see them?
The new animatronics are called Vilo Vele, Kado Bre'a, and Jibu Bre'a. They are based on the Anzellans alien characters from a Star Wars film released this year. Each has its own temperament and quirks, so the three robots react to the same thing in different ways. The first public showing will take place in Las Vegas at the end of this month, where event guests can talk with the characters in person. Disney acknowledges that the characters and the technology could eventually reach park attractions.
How it works without a puppeteer?
The recent Olaf robot, which The Verge mentions, is brought to life by a person: an operator can control it even from a portable console such as the Steam Deck. The Anzellans have no operator. Each robot hides cameras and microphones that capture guests' movements and speech. The data goes to AWS, where several AI models process it at once.
What are the key details on AI chooses, but does not compose?
The most important detail: the Anzellans do not invent speech and movements from scratch. Every line and every movement that the models select and trigger was written in advance by Imagineers, Disney's engineers and animators. This keeps the characters true to their personas, including each one's individual traits. As The Verge reports, AI helps guide behavior, but the Anzellans never deviate from the script. For a brand of Disney's scale, this is a sensible compromise.
Ship AI Faster with Founders, Engineers & Weekly Architecture Reviews
Inside Vibe Club (hamidun.com/club) — weekly live masterminds with Zhemal Khamidun, Tech Radar, full LMS access, and 370+ production Claude Code skills.
The AI world, distilled — once a week
Seven stories that actually mattered, hand-picked. No noise, no reposts, no press releases.
Done! Check your inbox for a confirmation.