Hello everyone,
I want to share my feedback on the Rabbit R1, which I’ve been using for three days now. My opinion is currently quite divided.
Hardware
First, let’s talk about the hardware: The device basically has everything it needs. The build quality is good, the buttons work flawlessly, and the user interface (I’ve included this under “hardware” due to the “external experience”) is attractively designed. However, there are some minor weaknesses here:
- Camera: The camera could be better that the AI can see more details. (especially near shots don’t work at all)
- Microphone: The microphone could be higher resolution. It’s not entirely clear whether this is due to the hardware or the software that records the memos.
Software
Here comes the biggest criticism: the software. At the moment, it’s unfortunately still a bit of a disaster. It lacks basic functionality and integration.
Connected Services
So far, I’ve only successfully connected Spotify and Midjourney. There was an error with Suno, but that’s it. ALL important integrations that would make everyday life easier are missing, for example:
- Calendar
- Contacts
- Tasks
Without these functions, the device can’t really be used as an assistant. These integrations are essential for daily life.
Follow-Up Messages
Another major drawback is that follow-up messages are not possible. This means that after asking an image-related question, you can’t simply ask another one without photographing the image again. This severely limits its use. Moreover, the depth of the answers is very limited. For example, if you ask for a recipe, it says it found something. But then it could at least provide the contents of the page. Currently, the query often ends at this “/sarcasm on” depth.
Rabbit Hole
The Rabbit Hole supposedly centralizes all queries and activities, but the device itself doesn’t offer access to it. The only interface is the browser-based Rabbit Hole function, which further limits the device’s application possibilities and integration into one’s own work environment.
Access to Previous Queries and RAG
It would be helpful if the device could access previous queries. Currently, this is not possible, which greatly limits the usefulness of the Rabbit R1.
A RAG would be helpful that vectorizes the Rabbit-Hole (that is the way that AI can remember, with bigger context than it can natively handle, which is important for scale).
Hands Free Mode
It would be nice to have a hands free mode, which is triggered via push to talk. That could be good when you use it on the desk while working. A custom triggerword would be helpful as well.
Vision Model
The vision model used by the Rabbit R1 is significantly worse than, for example, OpenAI’s. When I have an image analyzed, I get a more or less good answer, but can’t do anything further with it.
An example: My self-built AI assistant at home recognizes handwritten notes, digitizes them, and stores them in OneNote. The content is then vectorized and becomes “system knowledge” which I can query when needed. I would like a similar function here.
API Interfaces
Another important point is the lack of openness in the API interfaces. If these were at least opened up, I could build the assistance systems I expect or desire myself. Currently, however, the system is absolutely closed and not integrable for someone like me who develops such applications as a hobby. This is a major obstacle to realizing the full potential of the Rabbit R1.
It would be nice if all input signals (i.e., text, memo, image) could be connected in the API interface, and the output signals as well. For the audio output signal, the division between text-to-audio and direct audio would be very helpful. This would allow seamlessly incorporating the device’s response with the original voice or opting for a different voice.
Conclusion
I want to emphasize that I don’t find the device fundamentally bad. I like the aesthetics and the build quality. But on the software side, a lot still needs to be improved to make the purchase worthwhile. Features like follow-up questions and comprehensive integration should be standard these days. Especially opening up the API interfaces would be a big step in the right direction.
I hope that future updates will address these weaknesses and allow the Rabbit R1 to reach its full potential.
Best regards,
Zytruz
