The Rabbit R1 personal assistant, unveiled at CES 2024 by the emerging AI startup Rabbit, is a captivatingly futuristic AI device.
First things first, the Rabbit R1 doesn’t live in your phone or a home speaker – it’s a standalone gadget.
Co-created with the Swedish design firm Teenage Engineering, the R1 is a red-orange, pocket-sized square, roughly the dimensions of a stack of Post-It notes. It’s sleek, with a 2.88-inch touchscreen and an analog scroll wheel.
Atop the wheel sits the “Rabbit Eye,” a 360-degree rotating camera that points away when it’s not in use, serving as a default privacy shutter.
— rabbit inc. (@rabbit_hmi) January 9, 2024
Jesse Lyu, the brains behind Rabbit, envisions a world without fumbling through apps to complete tasks.
“We’ve come to a point where we have hundreds of apps on our smartphones with complicated UX designs that don’t talk to each other,” said Rabbit founder and CEO Jesse Jyu.
“As a result, end users are frustrated with their devices and are often getting lost. Rabbit is now building towards an intuitive app-free experience with the power of AI.
He continued, “Large Language Models, like ChatGPT, showed the possibility of understanding natural language with AI; our Large Action Model takes it one step further: it doesn’t just generate text in response to human input – it generates actions on behalf of users to help us get things done.”
The R1’s operation is intended to be refreshingly straightforward: press and hold the push-to-talk button, voice your command, and let the device do the rest.
Whether you’re summoning an Uber, making a dinner reservation, or adding a tune to your Spotify playlist, the R1 touts to simplify it into a single voice command.
But how does it accomplish these tasks without traditional apps or APIs? The secret is the Rabbit OS, which operates as a layer enabling access to various services through a web portal named the Rabbit Hole.
Users log into their accounts on this portal, granting the R1 the ability to perform actions on their behalf.
Moreover, the R1’s microphone activates only when you engage the push-to-talk button. This ensures that the device isn’t constantly eavesdropping like our smartphones probably are, a subtle nod to user privacy.
Underneath the interface lies a combination of large language models (LLMs) and Rabbit’s own large action models (LAMs).
These LAMs learn by watching humans perform tasks across various interfaces and then replicate these actions.
This learning process is what enables the R1 to perform a myriad of tasks, evolving and growing its capabilities over time.
An intriguing aspect of the R1 is its experimental “teach mode,” allowing users to train the device in new tasks. Imagine showing your R1 how to edit a photo or play a specific game and then having it replicate those actions independently.
The developers even used the R1 to perform simple functions in Diablo IV.
Priced at a pretty reasonable $199, the Rabbit R1 is positioned as an accessory rather than a replacement for a smartphone.
With pre-orders now open and shipping slated for late March, the R1 is intent on changing how we interact with technology.
It’s definitely an intriguing device at a reasonable price – but what will its true utility be? That’s what these first generations of AI devices have to prove.
Rabbit R1 feature round-up
Here’s a brief round-up of everything we know about the Rabbit R1:
- Device size and display: The R1 is about half the size of an iPhone, making it pocket-friendly. It features a 2.88-inch touchscreen.
- Processor and storage: Under the hood, the R1 is powered by a 2.3GHz MediaTek processor, paired with 4GB of memory and 128GB of storage.
- Rotating camera: One of its most distinctive features is the “360-degree rotational eye,” a camera that can be spun to face the user or away for privacy.
- User interaction: Interaction with the device is primarily through a “Push-to-Talk” button, invoking Rabbit OS to start listening and acting on user commands.
- Connectivity options: For connectivity, the R1 includes a 4G LTE SIM card slot, ensuring constant internet access. It also supports Wi-Fi connections and charges via USB-C.
- Battery life: Rabbit claims the R1 has an “all-day” battery life.
- Large Action Model (LAM): The R1 operates on Rabbit OS, utilizing what is called the Large Action Model (LAM). This model is designed to interact with and execute tasks on various user interfaces instead of relying on traditional APIs or apps.
- Capabilities: The device can handle a variety of tasks, from booking an Uber to suggesting recipes based on refrigerator contents. This versatility is demonstrated by its ability to learn and replicate actions, as shown in a demo where it learned to generate an image using Midjourney via Discord.
- Rabbit Hole Portal: The R1 features a dedicated training mode for teaching it new tasks, and users can manage services through a web portal called Rabbit Hole.
The Rabbit R1’s launch marks an exciting development in AI technology, offering a blend of aesthetic appeal, functional versatility, and user-centric design. Will it be useful? It certainly has the potential to be, but time will tell.
Its built-in privacy design and product messaging could be a subtle finger up to the Humane AI Pin, which was criticized for positioning its camera directly to survey your daily activities.