Opposable thumbs have given our species a pretty big upper hand In the evolution of life.
Starting with the ability to use tools, to write and record, these marvels of biological engineering have allowed us to develop multi-dimensional cultures and civilisations. (Millions of years later, of course, they’re being put to very good use to scroll, tap, and swipe across our phone screens!)
From the Humane AI Pin, to a rumoured collaboration between Sam Altman and Jony Ive, the hype around new AI devices replacing traditional smartphones is at an all time high. New entrant Rabbit R1 took this to another level when it launched at CES, a few weeks ago.
What is the Rabbit R1 And Why Should You Care?
The Rabbit R1 is a $199, AI-powered device that hit 50,000 preorders just 5 days after its unveiling.
So far, all the devices that want to be AI assistants follow instructions and do things we can do today by tapping our screens. Currently, the devices that are being touted as replacements for our phones can understand us, provide us with information, but not act for us.
That’s where Rabbit does things differently.
For example, imagine you want to book a table at a restaurant using Zomato. Currently you need to tap through multiple screens to do it. The R1 on the other hand, will execute a simple voice command and replicate what you would do, until the table is booked at the time and restaurant you want.
How does it do this?
The R1 uses a foundational new model called LAM or Large Action Model to train ‘rabbits’ or AI assistants that can do your bidding across apps and websites. This is in contrast to the Large Language Models or LLMs that chatbots like ChatGPT and Bard are built on.
But how does it work across apps?
Typically, for two entities to speak to each other, they would need access to each other’s APIs. For example, Siri would need access to Zomato’s API to be able to execute a voice command for booking a table. But it would be impossible to do this for every app – and that’s why it hasn’t been done so far.
The R1 hops over this hurdle completely, providing an app-free experience.
Each time you need to do something, you fumble through folders on your phone, opening apps and tapping multiple screens. The smartphone was supposed to be intuitive, but with hundreds of apps that do not work together, it no longer is. Our phones have become devices to kill time, instead of saving it.Jesse Lye, Founder, Rabbit.
As another example, consider the use case of booking a vacation. You could easily use ChatGPT or a similar LLM to plan your vacation and generate an itinerary, suggest flights, hotels, etc.
But ChatGPT cannot book your flight tickets and hotels for you. Like booking with Zomato, you have to do this today by tapping multiple screens.
Where ChatGPT would stop at generating information, the R1 would understand your intention and context, get approval on a schedule and then actually book flights, hotels, restaurants and events as Rabbit Founder, Jesse Lye showed in his demo.
Train Your Rabbit
R1 can also be trained on new applications, via a teaching mode built into its OS. In the video above, Jesse demonstrates how he walks R1 through the steps to use the Midjourney bot on Discord to generate an image and once done, R1 is able to do the same action by itself.
Going back to the Zomato example, if you taught a rabbit how to book a table at restaurant A, and next week you asked it book a table at restaurant B, at a different time and with a different number of people, it would understand you and adapt the steps it has learnt, to fulfil this new request.
It doesn’t stop there.
You can also teach the R1 a workflow that involves multiple apps. Imagine booking a table, adding the date and time to your calendar, and messaging invitees with the location of the restaurant. All without lifting a finger.
Sounds almost too good to be true, doesn’t it?
The bright, orange rectangle with its small touchscreen, scroll wheel, rotating camera, and a push-to-talk button on the side, has definite Gameboy vibes (Who’s old enough to remember those?)
The toy-like feel is exactly what Jesse, alongside design partner and long-term collaborator, Teenage Engineering, wanted to go for. The retro-futuristic style, the focus on tactility, and the playful look and feel are straight from Teenage Engineering’s playbook, simultaneously injecting a sense of nostalgia and newness – something that is sorely missing from our finessed, machined rectangles of metal.
Why a Device?
By this time, you might be wondering, “Isn’t most of the groundbreaking stuff in the R1 just software? Why do we need another device?”
It’s a fair question. Hardware is extremely hard to crack, because you are not only asking people to adapt to a new process, but also to adopt a new device that can take up precious pocket space.
However, both the Rabbit and Teenage Engineering teams were very clear that they wanted to build a product that formed an emotional connection and genuinely excited the user. In Jesse’s words, “It should be exactly like not a phone.”
Will Rabbit Win the Race?
The R1 is truly revolutionary because it is a universally connected platform that is company agnostic. It offers a peek into an AI-led future where there are no more walled gardens like Apple – only massive playgrounds
Theoretically, it seems utopian.
But as we have seen in the past, big companies will catch up and integrate LAMs into their existing, and already widely used, hardware. Where will this whimsical pocket assistant be then?
Only time will tell. As Jesse says, “phones have been around for decades, but we’re just getting started with AI devices.”
Meanwhile, I can’t wait to get my hands on a Rabbit R1 and put my opposable thumbs to good use!