Hey, welcome 50+ new friends onboarding The Signal last week. Today I want to share my interview with my decade-long friend Jesse Lyu, founder and CEO of Rabbit, the first AI Hardwarepreorders of 10 million dollars in just 5 days.
When Rabbit R1 was shown at CES 2024, people all over the techies were talking about it, good or bad. It almost stole the show.
And then everything changed, it sold 50,000 units priced at $199 in just 5 days. That's $10M worth of pre-orders in less than a week as a new company.
TechCrunch says Rabbit R1 is 2024’s most exciting launch yet. Fast Company says Rabbit R1 is AI hardware’s first success story.
Then on 18th Jan, Perplexity AI announced a partnership with Rabbit:
Together, we are introducing real-time, precise answers to Rabbit R1, seamlessly powered by our cutting-edge PPLX online LLM API, free from any knowledge cutoff. Plus, for the first 100,000 Rabbit R1 purchases, we're including a free year of Perplexity Pro!
That is a win-win strategy. But many people still do not know what the hell Rabbit R1 is. From the Rabbit website, it is your pocket companion. Rabbit’s Series A investor Vinod Khosla has a better saying:
People view Rabbit R1 as "hardware". To me a new class of device will enable a new user experience & new model for how apps are used, making it "AI agent" based user centric, not app centric. A large change is coming in very low latency (not achievable for apps on an iPhone) real time voice conversation based user interfaces. One app to use all other apps. See my prediction below:
Fast Company today published a story about Rabbit’s R1, the key is Jesse’s answer to why not make a phone or an app:
The R1 is not meant to replace a phone but rather be a new gadget that works alongside it.
Hardware offers more defensibility for his business compared to just offering an app.
The R1's key processing happens in the cloud, saving on hardware costs, and users can create and sell their own AI "rabbits" on the Rabbit OS platform.
To me, the key is Rabbit OS. In my interview with founder and CEO Jesse Lyu 6 months ago, Jesse told me what he finally wants to do is AI for OS: The next Operating System for AGI. And that is why Vinod Khosla invested in him:
There are so many things that this can go wrong, but if it went right, it's gonna be an empire.
Here is his plan in detail and we talked about these topics:
What might the next-generation OS for AGI look like?
Compare it to a PC or mobile OS, what could be the key feature?
Will OS be open-source or not, and how to build its ecosystem?
What are the biggest challenges? And what is ready?
What might the next-generation OS for AGI look like?
In my view, the evolution of OS can be summarized in two aspects:
Input/output communication between users and machines;
User interface and interaction;
Firstly, Let's look at input/output communication between users and machines
Computers were originally designed as non-intuitive tools. Intuitive tools (like a hammer, you pick it up and can nail things) do not process or compute any information, data or logic themselves. The output of intuitive tools depends only on external forces and physical laws, while computer operating systems themselves are non-intuitive.
They accept user inputs (text, voice, clicks, inputs, drags, etc.) perform computations at the chip and code level and then output the results to users in the form of sound, text, graphics, etc.
If we only focus on the input/output communication between users and machines, the traditional operating system industry mainly focuses on three elements:
Speed and efficiency
Ease of access
Data storage and synchronization
All of the above elements are relatively easy to understand. However, I often believe that whether an operating system can become a decisive product often depends on the front-end user interface and user interaction.
Secondly, the user interface and interaction
The output of computers is indirect, it is the result of calculations, processing and reasoning based on algorithms and semantics. The human brain was not designed to read binary code, naturally we are more receptive to language, graphics and sound.
Reviewing the progress of the user interface/user experience, we can observe an interesting pattern:
Punched cards - GUI (graphical user interface - holes can be seen as a graphical pattern)
DOS - LUI (language user interface)
Windows/Mac OS/iOS/Android - GUI
ChatGPT - LUI
It seems that every few years, the dominant operating system will alternate between GUI and LUI. To truly understand the backend logic behind the interesting pattern we discovered, we can examine it from the following aspects:
Humans interact with operating systems in an increasingly intuitive way;
Ultimately, most end users no longer need to understand how they work;
In summary, based on these findings we can reasonably infer the characteristics of the next-generation operating system. Here are my general ideas: For humans, the most effective, frequent and intuitive input/output is through natural language.
The human brain programs tasks according to who, where, what, when and how.
If we examine these five elements carefully, who, where and when are already handled by modern devices. We no longer need to remember everyone's phone numbers or birthdays, and machines surpass us in any geographical location or navigation tasks. We always rely on tools for time management.
Therefore, the key to significantly improving the end-user experience or the next-generation operating system, will be the mechanism of "what" and "how".
Compared to a PC or mobile OS, what could be the key feature?
