Around 5,000 people of the 100,000 who bought the Rabbit R1 are using the device at any given moment, five months after the device launched. That’s straight from the mouth of Rabbit founder Jesse Lyu, who gave the number to Fast Company while explaining that the device had to launch before it was ready in order to beat big tech companies to the punch.

That’s a big fall from the moment AI gadgets were having earlier this year. There was tremendous hype around the R1 after its CES debut, and an air of mystery surrounded Humane’s AI Pin before its reveal. Both shipped without any of the futuristic grandeur that was promised. As The Verge’s David Pierce wrote in his review of the R1, “the whole thing just feels broken.” Maybe that’s why almost nobody is using the R1 or why, last month, Humane was taking back more AI Pins than it was selling.

We’ve written at length about whether standalone AI gadgets have an ideal form or if their future is just phones. The AI features we’re seeing on smartphones are already good enough that even when an AI gadget does what its creators say it will, it feels doomed. Still, Meta’s Ray-Ban glasses make a good case for at least one type of standalone, AI-forward device. And who knows, maybe Jony Ive’s OpenAI device will, too.

Rabbit’s do-everything-for-you “large action model” update, which would ostensibly let it log in to websites and do things like order plane tickets or dinner for you when you ask, is apparently coming out on October 1st. Will it inspire people to dust off their Rabbit R1s and carry one more thing in their pockets, just for AI? Given that Apple and Google are both promising local AI that knows what’s on your screen and can do things for you across multiple apps... well, that feels unlikely.

Correction, October 7th: Lyu was initially misquoted in Fast Company. He now tells The Verge he was discussing concurrent users, not daily users, as this story initially stated.