Comment by zahlman
5 days ago
Historically I had thought there was a pendulum swing between using local computing resources vs. having a dumb terminal to access something remotely.
But now instead of swinging back to local resources, apparently we're proposing to add a second layer of remote access (phone -> computer -> Claude servers).
In the last 5 years I pretty much fully migrated to my laptop being a terminal for other machines. I more use it like a local machine in HPC: web browsing, word processing, scripting. Anything serious is done remotely. But I also live in the terminal and so realistically what's the difference? 99% of the time the result is that I get to use a "big" computer without having to carry it around.
FWIW, I'm not a big fan of AI coding. I use AI (including LLMs) and I am an AI researcher, but the vibe coding just hasn't clicked despite constant efforts. I guess it can make more sense to do it if you're programming from your phone because while normally typing isn't the bottleneck it definitely is on the phone (or at least far less comfortable)
Same setup as mine, I have an OpenVPN server running in my router, and my main PC has wake-on-lan and a KVM as a backup to turn it on and off.
I have an old used Dell Latitude that I use as a pseudo thin client. I ssh into my PC, and everything just works.
I really like this setup because I only have one environment, so everything is there, and I don't have to install anything in the laptop
Yeah that's one of my favorite parts. Same about living in the terminal. I can be effective anywhere nearly instantly. I carry everything around in my dotfiles and keep it small (keep the .git folder small and don't add anything except text files)[0].
On that note, one thing I highly recommend to people is to add some visual clues to tell you which machine you're on. I use starship and have a few indicators but I also have some PS1 exports that I've used in the past or use in new tmp instances (I HIGHLY recommend also doing this for when you're using the root account). It can get confusing when you have different tabs on different machines and it is easy to mistake which one you're on.
[0] I also recommend keeping notes there if you like writing in markdown. Files are so tiny that it's worth having them. It's benefited me more times than I can count.
5 replies →
My desktop is 11 years old, but I still feel like it does so much that I wouldn't want any cloud services except for AI. (And there's no way this thing would handle a useful local model, but I'm also really not very enthused about the kind of data sharing involved in remote AI use.)
I mean the power of the work machine really depends on what your needs are. Definitely should adapt to whatever your needs are.
So if you have a setup like mine then it is fairly trivial to incorporate that (or anything else). Either way you'll need a machine that can do the local AI though. Either that is on your "work machine" or you run the AI on a separate machine. You could even rent a machine and as long as you add it to your Tailscale network then you're connected.
I strongly suggest having a workhorse machine and then let other devices be your terminal into it. Your terminals can be very cheap (or an old machine) or as suggested, your phone.
3 replies →
Maybe in the future we'll all have a "hub" in our homes that contains our data, but we'll shell out to the local datacenter for AI compute, while our actual interface will be a VR headset or tablet located with us, anywhere in the world.
Hey, come on, it could be better: you could have hundreds of employees venting directly to chat logs held by Microsoft detailing all your internal politics, planning, customer acquisition strategies, code, integrations desires, excel sheets, emails, and projects.
Nothing could possibly go wrong, those guys are always 100% trustworthy and reliable, contracts and NDAs with them are ironclad and easily enforceable… … o_o