Is anyone using AI offline or locally on their own hardware?

[Midnight Georgia]

Been trying different things and wanted to compare notes.

The honest question is whether it is changing how you work or just adding a step.

Any experience is useful, even if it is just to say you tried it and it was not worth it.

What do you reckon?

[codeberg]

Been trying different things and wanted to compare notes. The honest question is whether it is changing how you work or just adding a step.

There is a bit more to it than that I think. I have fixed more machines by doing less than by doing the obvious dramatic thing.

Post back with what you find and we can go from there

[Sophie83]

Been trying different things and wanted to compare notes. The honest question is whether it is changing how you work or just adding a

Good point. Appreciate the discussion.

The gap between what people claim about AI and what it actually does in practice is still wide

[Pixel Mark]

Ended up in the same place, yeah. Good luck with it.

The gap between what people claim about AI and what it actually does in practice is still wide

[StringTheory83]

That is a popular opinion but I think it is wrong. We will see how it plays out.

The energy cost of AI is a story that is not getting nearly enough attention.

[Louise84]

Exactly what I found. I only really trust word of mouth for games now, reviews are all over the place.

Can't really go wrong with it

[VoidSentinel]

I see it slightly differently. I find the most honest reactions come out a while after the initial response settles.

Glad this came up.

The energy cost of AI is a story that is not getting nearly enough attention

[Nina81]

That is a popular opinion but I think it is wrong. We will see how it plays out. The energy cost of AI is a story that is not getting nearly

Good point. Good to hear other people's experience

[StoneCold]

Been trying different things and wanted to compare notes. The honest question is whether it is changing how you work or just adding a step.

That lines up with what I have been seeing. Let us know how it goes

[Jeffy]

Been trying different things and wanted to compare notes. The honest question is whether it is changing how you work or just adding a

Same here tbh. Might go back to it

[PowerhouseHobbs_Fan]

Started running local models after getting tired of sending every random thought to a cloud service. Nothing dramatic, mostly things like summarising my own notes, cleaning up documents, and experimenting with prompts. The biggest surprise was not the intelligence of the model, it was the convenience of having something available instantly.

The downside is that setup is definitely not as simple as opening a website. Drivers, model sizes, storage space, and settings can turn a quick experiment into a weekend project.

[ProperJobs]

Used a local AI setup for coding help and it has been a mixed bag. A smaller model running on my own machine is great for quick explanations and boilerplate, but for complicated architecture questions the bigger cloud models still have an advantage.

The nice part is privacy. Working with private code or internal documents without uploading them somewhere else is a big reason I keep it around.

[KeyboardWarrior]

A few months ago I built a small local assistant using Ollama and a couple of open models. The first week was mostly me saying "look what my computer can do" to anyone who would listen.

After the novelty wore off, the useful bits were simple things like searching personal notes, summarising long PDFs, and drafting rough ideas. It became less of a magic robot and more of a really handy tool.

[NatureBoy86]

Local AI makes the most sense for specific jobs rather than trying to replace everything. A lightweight model on a laptop is not going to beat a huge cloud model at every task, and pretending otherwise is just marketing noise.

For privacy, offline access, and tinkering though, it is brilliant. Having something that works during an internet outage is a surprisingly nice feeling.

[Shane_8]

Tried running models locally on an older gaming PC and the experience was interesting. The GPU mattered far more than expected, and suddenly that graphics card bought for games had another career.

The biggest limitation was memory. Some of the larger models technically ran, but the speed was closer to waiting for a kettle to boil than having a conversation.

[Phil]

The privacy angle is what convinced me. There are plenty of things I do not mind sending to a cloud service, but personal documents, work notes, and private projects are a different story.

Local AI gives a bit more control. It is not perfect, but knowing exactly where the data goes is valuable.

[Merchant89]

Not using local AI yet, but have been watching the space closely. The hardware requirements are still the barrier for casual users.

When phones and normal laptops can run useful models without needing a specialist setup, that is probably when it will become mainstream. Until then, it feels a bit like the early home server days.

[LogicWitch]

One thing people underestimate is maintenance. A cloud chatbot updates itself and you just use it. Local setups require keeping models, libraries, and drivers working together.

That said, there is something satisfying about fixing your own setup and knowing the whole thing belongs to you. A little computer hobby with a practical purpose.