https://web.archive.org/web/20240503092551/https://protos.com/creator-of-rabbit-ai-assistant-has-hidden-nft-past/

r1 founder's (scrubbed) past as another nail in Rabbit's coffin. I guess the 'large action model' is vaporware, back to Automator :D

BBC highly critical of Humane AI Pin, just like [[Humane AI Pin review not even close]] I noted earlier. Explicitly ties this to the expectations of [[rabbit — home]] too, which is a similar device. Issue here is I think similar to other devices like voice devices in your home. Not smart enough at the edge, too generic to be of use as [[small band AI personal assistant]] leading to using it for at most 2 or 3 very basic things (weather forecast, time, start playlist usually, and at home perhaps switching on a light), that don't justify the price tag .

https://web.archive.org/web/20240424051714/https://www.rabbit.tech/research

Seems the basic idea is: - Have a person perform a task (much like Automation on a laptop) and then the Rabbit (from taskrabbit no doubt) will be able to do variations on it. Automating in the style of [[Standard operating procedures met parameters 20200820202042]] - the device is a relatively simple edge , with most of the compute depending on data centers. This runs counter to smart edges, and interestingly counter to how they market the device (as no subscriptions or APIs needed). The simple edge should make it affordable and the centralised compute should make it scalable.

https://web.archive.org/web/20240424050235/https://www.rabbit.tech/rabbit-os

RabbitR1 claims about their workings allowing user actions not just information.

The Rabbit people say their LAM is a new type of foundation model, to be able to deduce user intention and decided on actions. Sounds like the cli tool I tried, but cutting human out of the loop to approve certain steps. Need to see their research what they mean by 'new foundation model'