A new device combines AI and voice commands with a new, universal interface which may take the clunkiness out of using smartphone apps.
The annual Computer Electronics Show was back to its sprawling, crowded worst this month in Las Vegas, drawing 130,000 visitors after several years of Covid-induced cancellations and reduced attendance.
I used to attend the tech fest on a semi-annual basis, given it was the go-to place for new trend-setting technology to debut. But years before Covid hit, the show lost its lustre. The crowds made the Las Vegas convention centre and surrounding hotels a nightmare to navigate. But more importantly, the show fell into a pattern of bland incrementalism.
Yes, there was always a new generation of skinnier, higher-resolution TVs and faster computers to be unveiled. In later years, electric cars and drones offered some wow factor, but products often remained beyond the reach and interest of the masses.
However, this year’s show featured one new product that points the way for mass market technology. The R1 device from Los Angeles-based AI startup Rabbit aims to tackle the big problem with that most ubiquitous of consumer electronic devices: the smartphone.
The smartphone app revolution that began 15 years ago has kept us glued to our phones but also created a complicated ecosystem of apps that need to be navigated in the course of a day - from Uber and Gmail, to the Calendar app, MetService, Spotify and Youtube. Voice assistants on the phone eased the situation to some degree by interacting with certain apps, but they generally only retrieve information rather than allow you to complete app-based tasks.
As Rabbit AI CEO and founder Jesse Lyu put it in a video timed to coincide with the R1′s debut at CES: “The smartphone was supposed to be intuitive. But with hundreds of apps on your smartphone today that don’t talk to each other, it no longer is. Our mission is to create the simplest computer, something so intuitive that you don’t need to learn how to use it.”
It became clear last year that generative artificial intelligence, based on a new collection of powerful large language models (LLMs), could be the answer in the quest for simplicity. ChatGPT effortlessly gave us answers to complex questions and application programming interfaces (APIs) began to allow applications to connect to large language models to deliver more specific and relevant answers.
The next era is one of AI agents - using AI to roam across apps and complete tasks on our behalf. But to do that seamlessly, we need a universal interface to make the job easy. Rabbit’s splash at the Computer Electronics Show is the first convincing vision of what that future looks like.
The start-up’s AI researchers have created what Lyu calls a “large action model”. It performs ChatGPT-like tasks such as answering general knowledge questions and assembling a suggested travel itinerary. But it goes further, allowing you to complete tasks, such as ordering an Uber or editing a document, all with voice prompts.
To make it easy to use, Rabbit made a huge call - to develop its own hardware. The R1, designed for Rabbit by Swedish company Teenage Engineering, is a square, slightly retro-looking gadget with 4GB (gigabytes) of memory, 128GB of storage and a 2.3 megahertz computer processor. The R1 has 4G mobile connectivity.
Pocket companion, not phone replacement
It has a press button on its side to trigger voice recognition, an analogue scroll wheel for navigating its colour screen, and a 360-degree rotating camera. Shake the R1 and a keyboard pops up on the screen. It has a built-in language translator that Lyu claims detects the languages being spoken and translates between them - though he didn’t demo this feature.
“Why would I need a new device if I already have a thousand-dollar iPhone?” is the obvious question Lyu addresses in the RI’s launch video.
“We did not build Rabbit R1 to replace your phone,” he says, “It’s just a different generation of devices.”
The R1′s screen features a series of tiles that display information based on the voice command. In Lyu’s demonstration, he asked Rabbit to book flights on Expedia for a family trip, played Spotify songs and used the camera to identify the contents of his fridge. Rabbit then came back with a recipe for a salad.
Existing services can do most of these things. But putting them together in one easy-to-use, voice-activated operating system (OS) interface is Rabbit’s real point of difference. AI will only achieve mass adoption when it is incredibly simple to use. That involves the R1 being able to integrate into the hundreds of apps we regularly use but allowing all sorts of tasks to be completed in one place, rather than hundreds of different app interfaces.
The early demos and reporters’ experiences of using the device suggest Rabbit has nailed the interface issue. The question now is how complex the task completion can get. Lyu’s demo also showed him editing a spreadsheet in Google Docs using voice commands delivered via the R1. In another demo, he issued voice prompts to generate images of rabbits in Midjourney, the AI-powered image generator.
Rabbit is now experimenting with Teach Mode, using machine learning to learn from your interaction with apps to automate processes on your behalf. That’s the first hint of how AI will be able to complete complicated, multi-step tasks from beginning to end.
The R1 goes on sale in around a dozen countries in March priced at US$199. There’s no subscription required to use the R1. That makes it significantly cheaper than other AI-powered hardware devices that appeared last year, such as the Ray-Ban glasses (US$299) and the US$699 AI Pin from Humane, which requires a US$24 per month subscription. The first run of 10,000 pre-ordered Rabbits has already sold out. New Zealand is not among the first countries to gain access to the devices.
I suspect that what Rabbit has created with its Rabbit OS and AI-powered R1 is already in development in the labs of Apple, Samsung and Google, which are best-placed to bring AI to the masses via our smartphones. But Rabbit’s vision delivers a fresh take on how we should interact with apps. AI agents have been touted as the way to cut through the complexity of the app economy.
“We are yet to produce an agent simply as good as users clicking the buttons,” Lyu reminds us.
The R1 may not survive long term as the “pocket companion” Lyu sees it as, but as a software interface and AI platform we use on our smartphones. Its success will depend on its ability to undertake tasks with voice prompts and simple screen tiles that you would typically have used your phone, tablet or computer to complete.
But it has served to focus the tech industry on what should matter - making the complex simple, as the iPhone so famously did on its debut in 2007. If Rabbit inspires tech companies to accelerate their efforts to simplify how we interact with AI, the bunny-themed startup will do us all a big favour.
