The AI box operates the APP for you. Is Rabbit R1 a magic weapon or useless?

Rabbit R1 is said to be driven by an AI large model system and sold 40,000 units within 6 days of its launch.

Written by: Mu Mu

At the 2024 CES show, a piece of AI hardware that can be put into a pocket became popular. The square small device called Rabbit R1 is only the size of a palm and can connect to WiFi, make calls, and log in to APPs on mobile phones. No operation is required, just shout, and Rabbit R1 will automatically navigate, call a taxi, and order takeout for you; a pair of “AI eyes” can also help you “see” unfinished business and handle work as needed.

It sounds like this hardware has similar functions to smart voice assistants such as Siri, Bixby, and Xiao Ai on mobile phones, but Rabbit R1 is said to be driven by an AI large-scale model system. Sold 40,000 units per day.

As a new smart hardware product, Rabbit R1 is positioned as an “intermediary between users and smartphones.” Company executives believe that this device can use AI to change the way users interact with applications.

As the sales of Rabbit R1 increase, there are also doubts: There are already smart phone brands that will add AI Agent (AI assistant) functions. Will people still need this “intermediary” at that time? The window left for Rabbit R1 is running out.

**40,000 Rabbit R1 units sold in 6 days. Where is God? **

When AI Agent became a specific application scenario for large models, Rabbit.tech, founded by former Raven founder and CEO Lu Chengxin, created the hardware AI assistant Rabbit R1. After this device was unveiled at CES 2024, it attracted not only attention, but There are also real sales.

After Rabbit R1 pre-orders opened, more than 10,000 devices were sold within 24 hours, far exceeding initial sales expectations. By January 16, Rabbit R1 opened four rounds of ordering, including 40,000 orders. Public information shows that Rabbit.tech raised another US$10 million in a recent round of financing, bringing the company’s total financing to US$30 million.

Rabbit R1 device sales are booming

Rabbit R1 uses WiFi or Bluetooth to interact with smartphones, laptops, tablets, smart watches, smart TVs and other devices. It can transfer files between devices and make voice or video calls. The built-in SIM card slot also allows it to connect mobile network.

Compared with other smart hardware, the biggest feature of Rabbit R1 is that it does not run any applications, but acts as a smart terminal assistant for users to use applications on these terminals to meet personalized needs. For example, when the user says “I want to take a taxi to the company”, Rabbit R1 can automatically open the online taxi-hailing software on the mobile phone and enter the destination to complete the user’s taxi-hailing request.

This is something that currently smartphone voice assistants cannot do. Siri can at most help you open the car-hailing app, but you have to do the specific operations yourself.

In addition to “hearing”, Rabbit R1 also has “vision”. There is a built-in camera based on computer vision technology on the back, which can rotate 360°. It not only supports taking photos, but also scans documents and converts them into PDF, text files or spreadsheets. And it can recognize objects, including faces, text, etc. in pictures or videos.

Rabbit R1 features a built-in camera powered by computer vision technology

The intelligence of Rabbit R1 comes from its Rabbit OS system, which is an operating system based on a large behavioral model (LAM). Compared with a large language model, LAM also requires a large amount of human operating behavior training. Users interact with the Rabbit OS system through natural language. After receiving instructions, the system can understand complex intentions and help users complete App operation tasks.

With continuous demonstrations, the Rabbit R1 system also has a strong learning ability. It only requires users to demonstrate the workflow once, and it can evolve through different interfaces and acquire new skills. At CES, Rabbit founder Lu Chengxin demonstrated how to teach Rabbit to learn to draw using Midjourney.

Indeed, there are more and more mobile phone and computer applications. Ordinary people may install two or three of them just for taxi-hailing apps. There are also multiple beauty APPs. Take e-commerce software as an example. Consumers who choose 3C products may fall in love with JD.com more. When buying clothes, they are more willing to visit Taobao. When looking for big-name alternatives, they have to turn to NetEase Selection. Each e-commerce company has its own main track and has a smart assistant. Helping you use the APP to meet your needs can indeed save a lot of time in clicking around, and even save money.

From the perspective of using applications to meet personalized needs, Rabbit R1 not only changes the way users interact with smart terminals and applications, but may also force applications to improve their user experience. In the future, when evaluating whether a software is useful or not, it will not only be judged by human users. Maybe some apps will be personally tested by AI assistants who understand code and machine language better and then be given a “bad review”.

**When large models enter the mobile phone, can R1 continue its sales myth? **

Able to listen and see, Rabbit R1’s outstanding features and sales undoubtedly add “potential” to the artificial intelligence hardware market.

In the past year of 2023, the conversational interaction capabilities based on AI voice assistants have given rise to a wave of emergence and even upgrades of consumer-grade smart hardware. According to iResearch’s calculations, the market size of my country’s core human-computer interaction products will reach 16.9 billion yuan in 2023. The statistical caliber of core product scale in this data is the number of orders signed in the market that year, including conversation robots, AI voice assistants in smart hardware, and voice AI chips.

Human-computer interaction product scale and driven output value

It can be seen that AI chips are developing towards being installed on smartphones, and some AI hardware has even set the flag to replace smartphones.

Previously, a wearable AI hardware called AI Pin has attracted widespread attention. This micro-projector that can be attached to clothes can project the operating interface into the user’s palm, allowing users to interact through gestures and voice to take photos. , translation and other functions. In addition, Ai Pin also has two breathing lights, which are used to remind users of incoming calls, text messages and other information. AI Pin also does not run third-party applications, but relies on large built-in artificial intelligence models to implement various functions.

AI Pin’s palm projection operation interface

The AI Pin, which sells for US$699 and has a monthly subscription fee of US$24 (including cellular network data fees and AI model usage fees), has the ambition to “change the way people interact with digital devices” and “replace traditional smartphones”. The device The founding team of the development company Humane includes former Apple design members Imran Chaudhry and Bethany Bongiorno. The couple has participated in the development of Apple Mac, iPod, iPad, iPhone and Apple Watch products and iOS systems. design.

However, the current AI Pin really does not show the ability to replace smartphones. In addition to its single function, the resolution of the green text interface projected on the palm is only 720p. It can be used to read text, forget about watching movies, and just play games. It’s even harder, you can only use one hand because the other hand is used as a screen…

Rabbit R1, which is also aiming to change the way humans interact with computers, doesn’t want to “replace” anyone. Founder Lu Chengxin said that Rabbit R1 is positioned as an accessory that provides a faster, more direct and more versatile way to integrate devices, saving people time wasted by various smartphone applications.

But the “accessory” Rabbit R1 is likely to be eliminated by future smartphones. Since the large artificial intelligence natural language model was pushed to ordinary people by ChatGPT, mobile phone brands have begun planning to install the large model into mobile phones.

In 2023, Huawei will be the first to integrate large models into mobile phones, allowing mobile phones to perform complex tasks such as text generation, knowledge search, data summary, and intelligent arrangement; Xiaomi will train a more lightweight large language model, and mobile phones will most likely be the first Carrier; Samsung is about to release the Galaxy S24 series of mobile phones, and Mingpai will integrate AI functions into Galaxy flagship mobile phones to provide users with end-side AI large model services.

Once mobile phones move towards large-scale models, at least the applications that come with the system will first iterate to meet the new AI functions. Other third-party applications will undoubtedly follow suit, and the smart voice assistant in the mobile phone does not rule out directly using the model to implement AI. Assistant function. Once this happens, Rabbit R1 may become a “redundant accessory” by “providing users with a complete universal controller suitable for various applications.”

It is reported that at 2 a.m. on January 18, Samsung will hold a global new product launch conference for Galaxy AI. At that time, the first AI phone expected by the outside world will be launched. I believe that after Samsung, more smartphones will line up to adopt AI.

It is said that Rabbit R1 will continue to be upgraded in the future, and AI phones are coming fiercely, and the window period left for this pocket box may not be long.