Our mission is to create the simplest computer. Something so intuitive that you don't even need to learn how to use it.
Last week at the Consumer Electronics Show in Los Angeles, Jesse Lyu (LinkedIn: Serial Entrepreneur, semi-pro Lamborghini Super Trofeo racer, music producer, car and vintage synth collector 💫) from a startup called Rabbit announced their debut computing device, designed to free us from the tyranny of apps.
The tyranny of apps
The pitch went something like this: When using a smartphone there are always endless buttons to click. Want a ride to the office? There’s an app for that. Want to buy groceries? There’s another app for that. The smartphone is a device that kills time rather than saving it.
Rabbit to the rescue
The solution? The company’s product is a little red tablet called Rabbit R1. It’s a thin client for a large AI model that runs in the cloud and interacts with apps on our behalf.
In the on-stage demo this looked pretty compelling: you tell it to order an Uber, it gets you an Uber. You want to plan holidays, it will take care of everything from planning to bookings.
The pitch went viral: within 2 weeks 40 000 people pre-ordered the device. Not bad for a product that doesn’t quite exist yet (first units expected in May).
The quest for a better AI assistant
Rabbit R1 is the latest contender in the long-running race to build a better AI assistant.
It started with the first-generation voice assistants of the 2010s: Siri, Alexa, Cortana and Google Assistant. These were working well for simple use cases (setting alarms, playing music) and most of them came with some way of integrating with 3rd party services.
But ultimately those systems fell short of human-level helpfulness: the couldn’t reliably handle complex tasks and take the right actions to accomplish them on our behalf.
Scalability
What we’re seeing today (between Rabbit and ChatGPT among other examples) is the second wave of AI assistants, powered by large AI models.
Even before LLMs it was already possible to develop an app that can handle some complex user queries. But it required custom logic tailored to each use case. An app that could handle wine pairings could not help with fixing grammar errors in an email draft.
Big models can learn how to break down complex queries into simple steps, and map them to actions that need to be performed on third party services. Importantly, they’re doing it at scale, without relying on humans to set up every flavor of every use-case that the model needs to handle.
Rabbit traps
There are good reasons to be sceptical about Rabbit R1:
💰 The business model is unclear. With 199 dollars per device and no subscription fees, how the company will pay for the infrastructure costs over the long run?
🔪 The space is very competitive, including ChatGPT and the older/ first-wave assistants, which will undoubtedly try to catch up. Importantly, the competition is following the easier way of deploying their assistants on existing devices
But what will ultimately matter most is just how helpful can Rabbit be. Their interaction model is all about browsing a 3rd party website, navigating it on behalf of the users, and explaining all meaningful choices back to them in short voice and text. It works in the staged demo, esp. when the “pizza ordering” demo is just “Order the most popular pizza”. But would it work for more complex use cases?
We’ll find out in May, when Rabbit R1 ships (if it ships 🤞). Until then, it’s best to think about it as a very cool experiment with much left to prove.
In other news
🥕 Rabbits (and hares) are my favorite animals. Fun fact: carrots are not actually healthy for rabbits. On the other hand, they are excellent for humans.
👀 Deja vu. Viv, a startup created by ex-Siri founders in 2017 had a very similar pitch to that of Rabbit R1 (article, video). We never saw how it works in practice: they got acquired before a public launch. (I wrote about Viv back in May 2022)
📈 We just reached 100 newsletter subscribers! Welcome to new readers 👋 .
Postcard from Bourg-Saint-Maurice
This week taking my first French skiing vacation in the Alps. Skiing seems to be serious business around here: a lot of expert skiers, and expert apres-ski party goers too 🥳.
Have a great week,
– Przemek
Pitbull, congrats on your 100 subscribers achievement 😃 🎉💫🌟
This is the thing I will never understand - why do people need an app to get to work or to buy their groceries...? Maybe instead of a new device they need a "real life" lesson? 🤔
Interesting! I doubt it'll be as smooth as how they pitch it but we will see