Andy Halliday

It works via your phone through iOS because this is an iOS expert who built that.

But it's not something that you can easily host yourself with a full LLM running locally.

But it is designed to do that.

You can do it if you have a killer machine.

But I asked...

I asked Perplexity, okay, look, I've got an M4 Pro with 48 gigabytes and a one terabyte drive.

Can I do a CloudBot installation on that and run a model locally?

They said, yes, you can, but you've got to stay down at the Lama kind of 7B level, Lama 3.3, for example, as the local open source model.

And if you tried to do anything large, you really need a 64 gigabyte memory machine to do that.

But Clonbot doesn't require you to run a local model.

You can just connect it to, it'll connect to any one of the models that you already have access to.

Interesting.

Yeah.

I didn't think about doing that.

help me and i don't know that i have a great answer on that but that's sort of where my brain goes with stories like this as as it does with other ones i start to float up in like two sentences capture you know what import of this is right so this is going this is going to be an ai that lives on your computer or

on a server in your home, basically.

Your computers are servers connected to that hub and so on, but that you command from your phone.

So it's this persistent memory for you and your whole family.

It's been designed in such a way that you can have multiple users in the same work group who have access to that with their own private partitions of the memory and

and data inside that so you can imagine i'm imagining now that i will use cloudbot or something similar i mean i i expect that ultimately anthropic will create claude family right they did claude co-work now they're doing a claude family and so you'll have an instance of cloud code and by the way this is not terribly different from cloud code with the exception of the fact that you interact with it largely by these phone messaging apps