The one good thing is that the AI features are opt in, not opt out.
I'd prefer to have it as an add-on or integration you have to manually install, but I guess I'll survive with inactive bloatware.
this post was submitted on 06 Aug 2025
66 points (93.4% liked)
homeassistant
15727 readers
4 users here now
Home Assistant is open source home automation that puts local control and privacy first.
Powered by a worldwide community of tinkerers and DIY enthusiasts.
Home Assistant can be self-installed on ProxMox, Raspberry Pi, or even purchased pre-installed: Home Assistant: Installation
Discussion of Home-Assistant adjacent topics is absolutely fine, within reason.
If you're not sure, DM @GreatAlbatross@feddit.uk
founded 2 years ago
MODERATORS
While I fully agree with the opt-in sentiment, 95% of all core integrations are likely "inactive bloatware" on your instance, yet I don't see anyone complaining about that.
Last time I checked we were at around 1500 core integrations on the search page.
It's not mentioned in the release notes anywhere, but the 2025.8 update has subtly changed the shade of blue used in the UI:
It's been throwing me off a bit ever since I updated.
Thanks for the heads up. I have an ESPHome device with a display where I've replicated the HA look and feel. Guess I'll have to update that now
They wanted that cornflower blue
I guess if I wanted AI in HA this would be how I'd want it done, but I don't really want it at all, so that's sort of a moot point.
Good thing fall is around the corner
The fall of AI?
Usually I wait for a .1 release before upgrading. But my 2025.7.4 hangs every few days anyway, so I assume it can only get better, right?
@thoralf
Wüst you should do is creating a second home assistant dev environment on your computer.
Doesn't matter if VM or Docker.
On VM you can make a snapshot.
Then backup your regular HA and import it on your dev.
In the dev HA you can check the update. If it does not work reset the snapshot.
If it works, do your update on the main HA.
As I didn't need it, unchecked, but maybe two HA could be installed on a raspi, 1 for live one for staging.
I always do a backup first and a snapshot on my Proxmox cluster. So, I can easily roll back in case the update is not working.
Still, some issues occur later down the line and then a rollback is still possible - but more painful.
Is AI in this ethical and sustainable AI?
...it allows users to configure their Ollama server or API key for OpenAI once, and then create many different agents using different models or configuration underneath...we’re also introducing a new integration, OpenRouter, which is a unified LLM interface giving access to over 400 extra LLM models.
I'm not familiar with these but sounds like they are leaving some choice to the admin
Is AI in this ethical and sustainable AI?
It's possible to make use of the full suite of HA AI tooling with entirely open source, on-premises, behind-the-firewall resources. That's actually damn awesome and can do some cool things. There's a lot you can do on a Raspberry Pi or an old Apple Silicon Mac, it doesn't take a lot of hardware for basic stuff. Ambitious usage is going to drive you to into a budget on par with "Gaming PC."
You can set up complex AI automation flows and voice assistant interactions without giving a penny to any of the many "ai" companies you might not want to support.
People are, in general, still boiling the oceans to produce all those interesting models you can run. Indirectly, you're sustaining that activity by giving it interest and inertia.
Where do you draw your line?
Does the streaming text to speech apply only to Piper running in home assistant cloud or does it include locally hosted Piper? The release notes aren't very clear on this.
@chaospatterns it's already in there since the last release:
> We have overhauled Home Assistant so our Text-to-Speech system can start generating the response audio before the full response is done generating. Last release we launched this for Piper, our local Text-to-Speech system.
Ah yeah you're totally right. I reread it a couple times and totally missed the "Last release we launched"