[pvtmert]

Does one really need to _buy_ a completely new desktop hardware (ie. mac mini) to _run_ a simple request/response program?

Excluding the fact that you can run LLMs via ollama or similar directly on the device, but that will not have a very good token/s speed as far as I can guess...

[ErneX]

You don’t, but for those who would like the agent to interact with Apple provided services like reminders and iMessage it works for that.

[claiir]

Oh this makes sense.

[titanomachy]

I’m pretty sure people are using them for local inference. Token rates can be acceptable if you max out the specs. If it was just the harness, they’d use a $20 raspberry pi instead.

[harveynick]

It is just for the harness. Using a Mac Mini gives you direct access to Apple services, but also means you can use AppleScript / Apple Events for automation. Being able to run a real (as in not-headless) browser unlocks a bunch of things which otherwise be blocked.

[znnajdla]

What other device would you suggest as a home server that a non tech person can set up themselves and has enough power to run several Chrome tabs? Access to iMessage is a plus. Small beeline Windows devices could also work but it’s Windows 11, slow as molasses.

[dplgk]

Their existing desktop or laptop computer?

[snigsnog]

Raspberry Pi using Pi OS

[fragmede]

You don't, that's just the most visible way to do it. Any other computer capable of running not-Claude code in a shell with a browser will do, but all the cool kids are buying mac's, don't you wanna be one of them?