Hacker News new | past | comments | ask | show | jobs | submit
I don't see any difference between the conversations around self-hosting local AI and self-hosting regular software, which have been going on for decades.

Yes, there are a hundred valid reasons to self-host. But 99%+ of consumers aren't going to care.

People don't even care when the cost of hosting is ~free, like a simple home server. How many are ready to buy thousands of dollars worth of AI hardware, which could very well be obsolete a couple of years from now?

The self-hosting crowd, Linux fans, OSS contributors, etc, are our own worst enemies.

If we care about self-hosting and open source and decentralization because we believe it leads to a better world, and not simply because we want to be control freaks and snobs, then we need to learn how to build shit that is actually convenient and usable and understandable to normal people.

I think it's possible. I think many people would want this, and care, if we could package it up in a digestible way.

loading story #50562689
loading story #50562683
Ollama and friends are pretty damn easy to use already. Hardware is the real barrier
let's see here, you already make a crucial mistake that a product manager wouldn't. (I say this as a senior software engineer that has worked with product managers a lot)

If I tell my mom to download and install Ollama on her Mac, do you think she's gonna have any clue which model to download from all of the 5000 models she can choose from? Do you think she understands which ones will even work well on her machine? What if she downloads one that's highly recommended online but it doesn't work on her 64 GB MacBook (hypothetical example)? Because it's too too large/slow...

and can Ollama, out of the box, do actual work on her computer, similar to the ChatGPT app with its "Work" mode, and dozens of built in connectors (google, etc)?

you call this easy to use for a normal person? You just made their point. The ChatGPT and Claude apps that my mom can download, set the standard for her. Her standard of good/easy is not what you and I as nerds find easy. you need to really put yourself in the shoes of a tech illiterate person. we haven't even touched on the fact that Claude and ChatGPT ship with easy to use iOS and Android apps that sync up perfectly with her Mac.

loading story #50562949
loading story #50574999
loading story #50563065
loading story #50563358
loading story #50562965
loading story #50562693
loading story #50567467
There's a decent opportunity where AI coding agents seem good at creating a very "average" and understandable UI. Maybe not perfect, but probably better than Gimp?
loading story #50562546
loading story #50563135
loading story #50563352
loading story #50562943
loading story #50575203
loading story #50564651
Hard disagree. I'm running Qwen3.8-Flash-Next using Strata on a Mac Studio and it covers all my needs. One drag and drop install. Nothing like self-hosting your email.
Seconding this. If you know how to build with a dumber model it scales crazy well. And Qwen 3.8 Flash Next is just wildly good even at low quants. I'm trying to find it's limits and I've been using it for over a week building and haven't hit a wall yet.
loading story #50562942
loading story #50564458
{"deleted":true,"id":50564438,"parent":50562753,"time":1795114307,"type":"comment"}
What are you disagreeing with exactly? Qwen 3.8 flash costs a few cents per million tokens from cloud providers. You can get effectively unlimited usage on a $6/month plan. You were okay with buying a $5000+ Mac Studio for it because you have the money, but you are in the tiny minority.
loading story #50564453
How much was your Mac Studio?
loading story #50564489
loading story #50563410
loading story #50563120
It’s more about self-hosting the hardware which is more desirable than thin clients.
loading story #50563195
loading story #50562842
loading story #50562486
loading story #50562684
loading story #50563205
loading story #50563041
{"deleted":true,"id":50562911,"parent":50562313,"time":1795108538,"type":"comment"}
loading story #50563099
loading story #50563750
loading story #50563177