Has anyone switched from areas to labels?

[WoJWoJ]

My light automations rely heavily on areas (I toggle all lights in an area on or off) because all my lights are neatly assigned to an area. Well, except that weird lamp that sits right at the border, and the CHO^(*) is never happy with it not being on or off with the right area.

My automations take that into account (because you cannot have more than one area assigned to an entity), at the cost of uglifying and complicating them. After some thinking, I realized

why using areas at all when I can have a label describing the place and have the lamp be both in the living room and in the kitchen?

Before breaking everything I wanted to ask if someone ditched areas in favour of labels, and whether this was a good or bad idea, and if it was bad why.

---

^{(*) Chief Home Officer}

[tom_l]

I’m pretty fastidious about assigning my entities to areas but I have not found a single use for them yet.

[WoJWoJ]

I use them to switch on an off groups of lights, especially in places where there are several smaller lights that only work well in unison

[tom_l]

Yeah I prefer scenes (scripts actually but same thing with more actions).

I use groups of door and window sensors for climate conditions and security but areas and labels only seem useful for control, not monitoring.

[WoJWoJ]

@tom_l

Yeah I prefer scenes (scripts actually but same thing with more actions).

I had a varied journey in my automation, starting with the first automation mechanism in HA, moving to AppDaemon, moving to my own code to automate, moving to n8n, moving to pyscript, and now back to automations

The thing that made me come back was the smooth integration of automations elsewhere in HA, as this is a HA “object” you find referenced here and there (as opposed to a script which is very powerful, but is not integrated)

I also like the concept of blueprints, which I will have a closer look at because some of my automations are repetitive.

Unfortunately I am constantly fighting my brain that says “fix it until you break it”

[NathanCu (Nathan Curtis)]

Labels do not equal areas in my install they are thier own thing… They act more as an index on a data lake. Primarily for llm use.

It turns into this :

Friday's Party: Creating a Private, Agentic AI using Voice Assistant tools

The Library Grand Index. The first avenge… tool. So I think you can see that tool use is going to become the big feature of LLMS in Home Assistant. In the first post, we discussed what is Agentic - and why should we design that way. Today let’s talk about what an LLM needs to operate your install, and how we give it those resources. LLMs work BEST when they have LOTS of context. (Yes, that’s the post about grandma: Issues getting a local LLM to tell me temperatures, humidity, etc - Configu…

So replace/ditch no

Augment and index. Yes.

[Edwin_D]

@WoJWoJ

smooth integration of automations elsewhere in HA, as this is a HA “object” you find referenced here and there (as opposed to a script which is very powerful, but is not integrated)

Home Assistant scripts are no less integrated than automations. Tom is not talking about Appdeamon scripts here, if that is what you were thinking. I have over 400 automations and over 250 scripts. Many automations call scripts instead of spelling out all the actions in the automation itself. It keeps my automations clean and prevents me from wanting to put too much in one automation just because the actions are the same.

I also prefer scripts (acting as scenes) over control by label or area. If I want to turn on lights in the living room, they are not all the lights, and each have their own brightness (and sometimes color). If I turn lights off in a room, I usually include lights that I’d normally not always turn on. It is only marginally more work to maintain, because scripts are reusable and I don’t add or remove lights in rooms every month.

[AshaiRey]

Area helps me to find thing in the right location and control that area. Labels are used here to group common tech related things like ‘voice’ , ‘ai’ or ’ security’

[WoJWoJ]

@Edwin_D

Home Assistant scripts are no less integrated than automations. Tom is not talking about Appdeamon scripts here, if that is what you were thinking.

No, as I mentioned, I wandered across AppDaemon, my own code, pyscript, scripts etc.

When I mentioned integration, I meant things like

Easily see the last execution and direct on/of switch

Directly write an automation for a device

All of these are of course doable with scripts and dashboards, I am referring here to the comfort more than anything else.

See, I am divided in this (and this is the reason for my various trials through the years) because I code so I like the Anything-As-Code approach, but weighing in the ease of integration for other family members.

A few more posts in this thread and I will fall back again into the “code” team

[Edwin_D]

Scripts also have a last executed timestamp just like automations. Scripts do not have an off switch, because the off switch disables the triggers. Scripts have no triggers - what would turning off mean? But the last thing you mention sends shivers down my spine.

That plus button for automations in the device is a direct invitation to use device triggers, conditions and automations. I avoid them like the plague. Besides that they do not even support the labels and areas you asked about, they are impossible to read - especially if the device is removed. Devices often have multiple entities: cramming all actions for them together is a jumbled mess. If a device breaks down all automations and scripts break with them. You cannot replace a broken device without fixing all references everywhere (which you can no longer read if the device is gone). Scripts not having the button is a blessing in disguise!

Why and how to avoid device_ids in automations and scripts

Background When you start writing automations, a device trigger seems the obvious choice. Everybody knows what a device is and it comes at the top of the UI dropdown list. [image] It works… but it is certainly not a “great way to start” because it is storing up problems for the future. Far better to use an entity, with state or numeric_state. This topic is part of the community-driven cookbook where you can find other topics you might be interested in or might like to contribute to. Devi…

LOL: I just checked the available triggers on my Skoda Enyaq device. I can scroll for days through the list

[Saoshen]

well said, I’ve also considered similarly;

areas are for physical grouping, labels are for logical grouping (in particular different device classes), groups are for same class