Yes, most likely your phone already has a virtual assistant in your pocket. You probably have at least one smart speaker in your house that you can ask to perform simple tasks if you’re reading Engadget. However, it appears that Rabbit, a recently founded firm, believes that Siri and Alexa are not the best examples of artificial intelligence (if you can call them that). This week, the business, which is exhibiting at CES 2024 in Las Vegas, imagines a world where you trade apps for conversation and communicate with what essentially functions as an AI walkie-talkie instead of a bothersome device that shoves icons in your face.
Launched as Rabbit’s debut gadget, the R1 is an undeniably cute tiny square in a charmingly vivid orange color. The visual attraction of the Teenage Engineering-designed device is undeniable, even if you’re not convinced that a specialized device is needed for a virtual assistant. It has two microphones, a speaker, an analog scroll wheel, a small 2.88-inch touchscreen, and a “360 degree rotational eye,” which is essentially a camera that can be rotated to face you or be seen through the back of the phone.
However, the “Push-to-Talk” button is the main button you push and hold to interact with the R1. Rabbit OS is informed to begin listening by this. As you ask it a question or assign it a task, a highly stylized and disembodied rabbit head slowly bobs and then promptly goes to work. Do you want to reserve an Uber ride? In need of a recipe to finish off the food in your refrigerator? Who sampled “That Lady” by The Isley Brothers, you ask? (Beastie Boys, Basement Jaxx, and Kendrick Lamar, FTR, are the answers.)
Using a concept it refers to as the Large Action Model (LAM), Rabbit OS is able to handle those duties. Jesse Lyu, the company’s founder and CEO, presents this as the main innovation. It is intended to be used with interfaces as opposed to apps or APIs for action taking. To put it briefly, it can be trained to perform nearly any task that can be completed using a user interface. It resembles a more elegant version of a macro.
Lyu uses Discord to educate the R1 how to create an image using Midjourney as a demonstration of its potential. When Lyu walks and completes the mission, Rabbit OS records his movements and can replay the task upon request.
As a kind of privacy shutter, the spinning camera is oriented by default to face up into the body. When called upon, it will only turn its sensor toward its target. It can perform standard tricks, such as recognizing objects or people in the actual world (within reasonable bounds, anyway). But people’s curiosity will be piqued by the way it interacts with the AI. During the demonstration, Lyu gestures toward a fully stocked refrigerator and asks the R1 to recommend a dish that is “low in calories” based on what’s inside.
Naturally, there are still a lot of unresolved issues with the R1. What is the battery life like? It’s “all day,” according to the corporation, but what exactly does that mean? And will it be simple enough for the typical user to train? But at least we have some knowledge. We are aware of the price, which is $199, and that it is already available for preorder. A March or April shipment date is anticipated.