Roadmap 2025: A Truly Smart Home through Collective Intelligence

agreed, I’d like to add and mention the one integration we all really need: Powercalc which has a device database growing by the day, powered by the community.

Btw, there is a lot of ambition (nice), and things we want in the above, but let’s for once also talk about what we don’t want.
An Anecdote of our Home Assistant system the other day:

Asking my voice assistant to turn-off some scene, I didn’t work and I got annoyed, telling it with some extra wording (‘nou eens’ in Dutch, probably something like ‘listen for once’ in English) my Assistant is setup to use Dutch language, but this time it replied: Ok, turning off scene X and The House… Which it did…

A smart home would have said, I am sorry I dont understand (ok, it does that all the time which is also annoying…), and it should have at least asked: “Are you sure you want me to turn of your House…?”

Bloopers of that kind are killing faith in any roadmap here in our household talking about Truly Smart…

As a matter of fact, Ive reviewed the exposed entities with calamities of this kind in mind once again. No more voice assistant exposed entities that can have such disastrous consequences.

I am now forced to go to the Dashboard again for those, or use their dedicated app circumventing Home Assistant altogether.

couldnt agree more, the presented Roadmap diagram seems not very well balanced. How can ‘making Music Assistant easier to setup’ be more important than Privacy/Security. Those elements aren’t even of the same level of abstraction so impossible to represent like this. As if Privacy would be a Dashboard item only (??? indeed).

I do hope that the new concept of Integrated devices wont result in even more indivual entities. Like with the previous move from integrated entities with several attributes.
Our systems are seeing more and more of those and overloading it further per release so it seems. If I compare my system with a year or 2 go, it must have trippled the numbers. Simply checking dev tools states takes practically half a minute before it has loaded…

Please consider that aspect also when introducing new models of context and information. Especially when they are additional, and not replacing current architecture.

Oh, totally forgot about this one: there is untapped financial potential for NC in this topic. At the moment when exposing HA to be accessed from outside your own network, it’s basically an “all or nothing” approach when using the default ways (NabuCasa, VPN, Wireguard, …): either my system is completely blocked to the outside world, or fully accessible. All devices, all logs, all automations, etcetc, everything. Including areas, where remote access doesn’t bring extra user value, but is a security risk when exposed to unauthorised people. For that exact reason I e.g. use a telegram bot as only access point from outside my network. Only what I have defined to be accessible from outside is also reachable.
If NC would offer a remote access approach that would limit things to e.g. a “read only” mode, or a “you can only switch things, not change configs and gain access to everything” mode, this would make it interesting to a user group that currently refrains from using remote access. How big that group is and if it’s a relevant market to target? That’s something to find out via surveys etc :slight_smile:

edit: okay, my curiousity won, I started a survey in the facebook group: Home Assistant | Hey everyone, especially the ones who are currently NOT having full access (via NabuCasa, VPN, wireguard, ...) to their HA system from outside their n... | Facebook

1 Like

When working on user/security, it would be great if users can be referenced by username, instead of obfuscating it by id (see current card visibility user condition). Username could be used on frontend as it’s readable, and it can also be used on backend - it’s already unique, so no need for another unique, autogenerated db field.

I would love to see HomeAssistant launch their own AI model specifically tuned for the smart home environment! Available for local install via Ollama but also in the cloud for Nabu Casa subscribers.

3 Likes

I believe Home Assistant should no longer be considered solely for home use. For example, I design smart office spaces in my country powered entirely by Home Assistant. One of the biggest challenges I face is not being able to safely expose the system to office employees.

A “Guest Mode” (or limited user access mode) is absolutely essential. For instance, users who log in as guests should not see the sidebar, or even the top bar. They should only be able to view their assigned dashboard, and nothing else.

In one project, I couldn’t give employees access to the smart door lock because once I exposed it to the interface, several unrelated settings also became visible to them — even though they were not admins. I had to resort to using a Telegram bot just to ensure secure, limited access.

A proper guest mode should be easy to configure and user-friendly. If this feature existed today, I could simply ask all office employees to install the Home Assistant app and log in with their personalized limited credentials, safely and confidently.

Another feature I strongly recommend is a notification history. Currently, once a mobile notification is shown, it disappears forever. There should be an option to store and review past notifications, at least for a limited time.

8 Likes

There’s a lot that I’m excited about from this post and I agree with a lot of the conversation (eg: energy dashboard - everything I wanted to say has been said already).

The three things I’m going to note:

Enhanced privacy controls for users,

This is why many of my household don’t interact with HA outside of the tablet or automations I’ve set up. I /want/ to give my kids control over their room light/fan. They should NOT have control over MY light/fan (or quite honestly the worst is control over their siblings devices). But separating that out is a bit of a mess right now and really needs to be addressed better.

I’d LOVE more insight into “what the heck just happened?” moments. For example. This morning. Automations were all working fine. Wife opens the door (with a door sensor on it which then triggers other automation) and … nothing happened. Huh. Close door and re-open because that z-wave door sensor triggers like 19/20 times for no reason I can determine. Nothing. Um. Go to kitchen HA tablet. Check HA. Turn on z-wave light - nothing. Turn on zigbee light - works. Turn on wifi light - works. Turn on another - z-wave light… nothing. Fine. I open HA on my device to look and before I can bring up any of the tools - ALL of the z-wave automations kick on almost simultaneously. I have just under 80 z-wave devices and sometimes they are just slow for no discernable reason. But anything that a) can tell me “what the heck just happened?” without digging through all the logs of human-jibberish-id codes would be amazing b) can tell me “hey, your z-wave network is slow because this one device over here is spamming the network” would be appreciated.

I love GitHub - andrew-codechimp/HA-Battery-Notes: A Home Assistant integration to provide battery notes of devices - like, seriously, everything about it. What I want but have never been able to replicate yet is something that simple for things I consider information I want to know about. I can set up alarms and things for like if the z-wave smoke detector goes off. But I gave my wife an over-ride for light automation - but then she’ll trigger the over-ride but forgets to turn off the light and so the bathroom light has been on for seven hours while we were out of the house (this is a weekly issue). Just a little pop-up display “bathroom light on $x hours”. The yard garage-door - the wind can absolutely open it if it isn’t bolted and the kids will forget when they put their bikes away. I don’t want alerts throughout the day when they are going in and out - just a simple status of the door lock on my home screen will do. I have SO MANY rules/alerts/ect for what amount to trivial bits of information I want to know but yet I don’t want boy-who-cried-wolf alerts on this trivial stuff. I’ve yet to figure out a way to do it that is as clean and simple as Battery Notes is for devices low on battery. This would go a LONG way on spousal approval too. Something simple she can check on her whim without being pestered by alerts.

And if I’m still just to newb and am missing something on the above because it already exists and I just don’t know - feel free to correct me. :laughing: (But I might also be the target of the whole point of the blog post in simplifying a lot of this too :rofl: )

Thanks!

Love that the team is planning on making the roadmap more public. THis helps us know what we’re going to see in the year but also help provide feedback (in whatever format that ends up being). For me that makes me feel a bit more invested in the project.

Great work and looking forward to the rest of 2025’s releases.

Last year when I saw the first roadmap I was very pleased. Its orientations and goals were in my opinion exactly what Home Assistant needed to follow to succeed among the other big tech home automation platforms.

This is again confirmed with this year’s second roadmap. I’m truly amazed (and relieved) by your acute understanding of the challenges for the future of open (source/standards) smart homes. Let’s face it, usually open source projects are made by geeks for geeks, technological beasts, but Home Assistant is really leaning into the direction of becoming a real general public product and that’s great news for everybody because it is the only real open contender. And open will always be the best option to both preserve individuals’ and common best interests.

I have for once high hopes in the future of the open smart home!

Great work team !

2 Likes

Could the energy dashboard support district heating? We have a lot of that in Finland. Hot water is produced and distributed centrally at city level. The unit of measurement for the energy is kWh.

2 Likes

Amazing work. Smart home administrators aren’t your average person on the street. Very tech savvy, with home automation becoming a hobby for many.

To include kids and wife, we need features like Oauth/openid so they don’t need another username and password… an authentik integration would do wonders.

Docker installation without supervisor is preferred choice, not haos.

Kids / tenants should be granted specific access to specific devices/rooms.

1 Like

An even worse experience yesterday makes all of this of secondary interest

As long as Home Assistant can not prevent (or stop causing) accidental calamities like that , any roadmap is futile.
Never happened before, but Murphy tells us it can so it will. And it did.

I do hope the above scenario is considered under the target of improving security.

Don’t self destruct…

Hold up. I’m not disagreeing per se, but I don’t understand how the smart home being managed primarily by one person has to do with making it easier to manage the smart home, or with administrators not having the complete picture of their system. The remedy for an administrator unaware of the pinch points in their home would seem to be a better logbook and traces (and perhaps making those “why did HA take this action here” sorts of logs more accessible to non-admins).

Unfortunately, there is still nothing on the roadmap in the direction of making usability better for assistive technologies users. Not enough people have disability, which is a good thing but let some people excluded because YAML is still the best way to configure automations for example, which is not really beginer friendly.

1 Like

It does make a lot of sense! Samse case here, I have “fancoils” for heating that are really computer fans with ESPHome controller + temperature sense tied to my radiators, I understand the combo for me is “fancoil”.

Getting here though depends on the complexities of the changes need, but it is interesting.

Yes we are! (Product person not dev sorry :+1: ). But thanks for sharing them either way.

1 Like

We want to walk there, its a long road but we need to start somewhere :smiley: Once we create the fundamentals there is a lot of power to unblock, the real challenge is to create the steps so we keep creating and evolving over time until the “too good to be true” goal. But we’re ready for the challenge :muscle:

This is a terrifying story. I wouldn’t want my house to do that.

2 Likes

From a TECHNICAL side I get you, but your scenario simplifies things from a “living a good life” perspective. A real life roommate can tell the difference between “he has a bad day, I’ll just do the dishes for him to help out” and “this is the 20th time in 2 months, there is something off behind the visible behaviour, let me talk with him and see what he is struggling with” scenarios. For a smarthome to achieve that, it would have to include a psychological aspect, and that’s where i also wouldnt want to go. Neither with current nor with future ai capabilities.

And to add another aspect that i think most of us tend to forget “by design”: when thinking of an “ideal smarthome”, where is the right limit in capabilities when it comes to children of that household growing up? Where should a smarthome stop offering things in order to not replace crucial learning and developing.

I’m 45, i know I’ve probably reached the limit of being able to keep up with remembering daily chores, so I’m fine with setting up a “take the f****** trash out!” reminder for me. Is it healthy to let a kid grow up in a home where it doesnt have to learn how to remember such things?
Think of a kid that grew up with a butler and a housemaid and never needed to learn how to boil water for tea.
Extreme example, i know, but just to get the principle across

An often undesirable side effect of making things really simple for new users, or for those with little technical background, is that it can limit the amount of control people have for some aspects of the system. A couple examples that are related to network traffic on the LAN:

  1. Subnet wide reverse DNS lookups every hour.
  2. Hard coded backup DNS.

Both of those items have had extensive community discussions about them, often contentious. But those two items do cause real world outside of HA itself problems for some users, primarily in the form of creating a lot of packet traffic noise on people’s networking monitoring tools. Which then makes it harder to spot the other things going on within the network that might actually need attention.

In the case of new device discovery, it is not needed to be there for 99.99% of the time that the system is up and running for many of us. It just clutters up the network with useless traffic.

So with that background, what I would like to see as part of this planned architectural work to have provisions for a more centralized “advanced user” control area to be able to selectively disable those kinds of lower level things. This would be conceptually similar to how web browsers typically have a means to get deeper into the configuration of the browser than normally accessible via menus.

Yes this will give people means to shoot themselves in the foot, so it would need to be made not too easy to get into, and need to put up a warning that this is potentially a danger zone, ‘so make sure you that know what you are doing’ before proceeding.

All too often the zeal to make things easier for the new user masses ends up tying the hands of those who have taken the time to understand the inner workings of the system.

The bumpy initial roll out of the reworked backups in 2025.1.x is another example of the conflict between seeking ease of use and simplicity vs. what the more advanced users actually need. Please don’t assume that everyone is incapable of understanding the inner workings of their networking environment, or other lower level aspect of their systems.

It is a balancing act, so I do get that the upsides are much about seeking to reduce support burden, and for generally less frustrations for new users. But that tradeoff for favoring ease of use over total user control isn’t always the best for everyone. And I believe with the proper design considerations from the outset, those need not be mutually exclusive.

3 Likes