Blog

If you’ve been online at all this week, you’ll have noticed a lot of reviews for the new Humane AI Pin and Rabbit R1, which are, frankly, brutal. Riding the AI ​​hype train wasn’t enough to mask the incredibly limited “under development” nature of their software or some completely broken functionality. The consensus is clear: Don’t buy one.

These products may even have deflated the inflated AI bubble, at least a little. Many of the top smartphone makers, who are also rushing to cram AI into every possible facet to turn their fortunes around, should also look to Humane and Rabbit as a cautionary tale. So are phone users.

Even when they work properly, part of the problem is that there isn’t an extremely compelling reason to use these standalone AI products. As we place it in our hands; it’s hard to answer why Rabbit R1 isn’t just an app (it turns out you can run it on your phone). Yes, voice commands that free you from the screen are great. Likewise, AI chat, text summarization, translations, note-taking/transcribing, and even real-time object detection can be useful, but they’re not game-changing enough to justify an entirely new product category.

You’ll find many of these same features advertised in newly created AI suites for smartphones, such as Samsung’s Galaxy AI. However, I have hardly used Circle to Search or voice recorder transcriptions since their introduction. Regardless of what companies may think, AI in itself is not a selling point, and users will quickly forget or tire of features that aren’t really useful. Numerous polls we’ve conducted also show that most of you aren’t that worried about AI’s generative features. Simply put, brands can’t rely on AI buzz, or lack thereof, to shift units.

Worse, though, is that these AI-centric devices offer very few third-party app integrations that could, in theory, make them truly useful tools. Spotify, Uber, and DoorDash, when their web-hacked integrations even work, are your part of the Rabbit R1, while Humane has Tidal for music, and that’s it. What a joke, even for early-stage products.

However, the mobile phone is no different. The Galaxy AI web page summary only works in the Samsung browser, but most of us probably use Chrome or Firefox. What if I want Google to aggregate email from a self-hosted provider or play media from local storage rather than a popular cloud service? This is the big problem that AI and even the latest generation of assistants have yet to solve. Google, Apple, Microsoft or whoever needs to figure out how to connect AI to the vast ecosystem of apps that users currently enjoy. Even more difficult will be convincing developers to go cross-platform, which proved incredibly difficult in the heyday of virtual assistants. Either way, reinventing the wheel with proprietary features won’t create an AI platform that consumers will want to invest in long-term.

This point brings us to the heart of the problem with the Rabbit R1, Humane AI Pin and other AI products on the market. They all rely on cloud computing to run the large language models (LLMs) that make up their core functionality. In fact, both seem to rely on OpenAI’s ChatGPT-4 model in at least some form, probably because it’s considered the cutting-edge LLM at the moment with a solid API behind it. These devices may be little more than a wrapper for third-party APIs, like many of the AI ​​apps you’ll find scattered around the Play Store and even the enterprise space. If ChatGPT crashes or your network suffers from performance issues, your Rabbit R1 becomes a paperweight. But this is also a problem because all integrations are only deep.

Building products that rely entirely on third-party services has always been pretty silly, as services can change features and costs or shut down altogether. How many times have we complained about outages in Google’s own ecosystem? However, the AI ​​arms race has led many companies to abandon this piece of common sense. Looking at smartphones, this highlights how important the work on Gemini Nano and Google’s Android NN API really is. Nano runs small models locally on the device, without the cloud, ensuring functionality is protected and works even when your phone is temporarily offline.

On-device AI requires sophisticated hardware and software that aren’t cheap to invest in or build (there’s a reason the R1 is only $199), and it still doesn’t automatically solve the application and use case problem. Still, if there’s an AI strategy that’s going to win, it’s almost certain to be one that takes time to figure out all those boring but necessary basics first.

So, let the Rabbit R1 and Humane AI Pin be a warning to those aboard the hype train and to smartphone brands (and consumers) with AI tunnel vision. It takes a lot more than just shiny-sounding AI features to make a truly compelling product, and we’re probably still a few years away from solving these early problems.