Roadmap 2025: A Truly Smart Home through Collective Intelligence

It’s been a year since we launched our first roadmap for Home Assistant, which means it’s time to launch this year’s first update! In 2025, we’re building on our direction to make Home Assistant easier to use for everyone in your household, and taking it to the next level by making smart homes more intuitive, proactive, and user-friendly. 🚀

This centers on helping Home Assistant understand devices in their context, and cataloging every device that works with Home Assistant through a new project, Device Database. And we will do this the Home Assistant way — through the power of our community and our collective intelligence. (…but that’s not all! The roadmap also covers many product areas we are aiming to work on from automation to voice to dashboard and more!)

If you’re new to our roadmaps or curious about how they guide the development of Home Assistant, be sure to check out our introduction to Home Assistant roadmaps. As always, your feedback is incredibly valuable, so please give us your thoughts in the comments.

One year on the road

Last year, we introduced our very first roadmap—and it wasn’t just plans on paper. It marked a new chapter for Home Assistant: one where we made our direction clear, set ambitious goals, and invited the whole community along for the ride.

In 2024, we focused on building a smart home that has a high Home Approval Factor: helping both the maintainer and residents of the smart home unlock more value from Home Assistant without needing to be an expert. This was done by improving the touchpoints that all members of the household will interact with, such as automations, dashboards, and voice interactions, while maintaining the power and depth of the platform for our power users and admins.

The real challenge lay in building a roadmap from scratch. That meant not only poring through and prioritizing every single feature request, but also increasing transparency and building trust in the roadmap within our organization and our community as a beacon that can help guide the development of the project. Thanks to the feedback, energy, and contributions from our global community, we’ve made huge strides. In the past year, we tackled some of the biggest pain points, such as drag-and-drop dashboard improvements, automation grouping, standard voice assistant hardware, better automated backups, and more.

Now, in 2025, we are ready to lay the groundwork for the next big step forward: Making Home Assistant truly smart through our collective effort and intelligence.

The smart home administrator as an inventor

Home Approval Factor means that a successful smart home isn’t just great for the person who set it up, but for everyone who lives there. Whether it’s your partner, kids, or roommates, we want Home Assistant to feel intuitive and supportive for the whole household.

In some ways, being a smart home administrator is like being an inventor (or, as we would say, a product manager). To be a great inventor, you need to understand your users. What are their needs? What solutions can you invent to solve their problems? How do you know you solved their problems? Can you improve upon the solution?

Home Assistant provides all the building blocks for our administrators to build whatever they want and let their imaginations run wild, and better yet, it’s easier than ever with the improvements we made last year! However, that’s also the curse of a blank slate. Inventing fun and hyper-personalized solutions for niche problems is a major appeal of Home Assistant, but sometimes, maintenance of a smart home can be a chore — not every single problem requires reinventing the wheel.

The reality today is that, according to our community survey, just 46% of partners and only 27% of children of long-time Home Assistant users are directly interacting with Home Assistant. That’s a big gap, and it highlights a broader challenge: Even the most experienced smart home administrators don’t always have the complete picture of their household’s needs. They may not realize what would help until something goes wrong. And it’s unreasonable to expect everyone to be an expert in everything — from automations to dashboards to onboarding new users.

From shared wisdom to collective intelligence

Right now, if a smart home admin wants to improve their setup, they have to go hunting—through our community forums, GitHub repos, Reddit threads, YouTube videos, Discord chats… you name it—trying to find tips and tricks that apply to their specific case. You share incredible wisdom with each other every day, but much of it is scattered, short-lived, or hard to apply without lots of customization.

We believe it’s time for a smarter system.

What if Home Assistant could learn from our community’s best ideas and use that knowledge to proactively suggest improvements? What if it could guide users based on what others have done in similar situations? That’s the vision we’re working toward: using collective intelligence to power a truly smart home.

To get there, Home Assistant needs to understand what’s in the home. Inside each home, there are the people, the spaces, and the devices. This year, we will start with helping Home Assistant understand devices in terms of (1) context: what each device is and how it is used, and (2) information: every device that works with Home Assistant and how it is supported.

Putting Devices in Context

Today, most devices in Home Assistant show up as a bundle of entities: a temperature sensor here, a switch there. This is by design — we want to provide the ultimate flexibility in configuring your system, and will continue to do so. However, it is not without disadvantages.

For example, we don’t always know that those entities together form, say, a fridge, which is essentially a mini house of its own: It may have temperature sensors, a door sensor, a light, and so on, but Home Assistant cannot assume them as any other sensors or lights within a room. Without that context, Home Assistant can’t do much beyond letting users build their own dashboards and automations from scratch.

As you add more devices, entities pile up, and Home Assistant starts to lose the thread of what’s actually in the home. Your smart home should become more powerful as you add more devices, but at the cost of the difficulty of maintaining them also increasing exponentially.

When we put devices in context, everything changes. A fridge becomes more than a list of entities — it becomes an actual device that can have a dedicated dashboard, default automations, and contextual voice commands.

Here’s how context unlocks a more streamlined experience across the platform:

  • Voice: Assist can currently source the right sensor if it knows the area of the device. In the future, we can also intelligently exclude irrelevant sensors from a voice query. For example, “What’s the temperature in the kitchen?” shouldn’t return the freezer’s internal temperature.
  • Dashboards: Integrations can provide customized cards and dashboards right from the start, providing a much more streamlined experience. For example, we can provide a fridge dashboard. It works with other devices, too. Imagine a car dashboard or a 3D printer dashboard.
  • Automations: By understanding what devices a user has and how they are commonly used, we can intelligently suggest automations created by our community. For example, “fridge door left open” or “fridge water leakage” alerts can be provided without building them manually from scratch.

This will be a core architectural change that will be proposed and discussed in the coming months. We’re building on what our contributors already share in integrations: knowledge about how devices are structured and how they should behave. My hope is that this will provide the framework for our contributors to contribute dashboards and automation blueprints to our code in exciting new ways.

Introducing the Device Database

To make all this work, we also need a centralized, structured place to store and share device knowledge. That’s where the Device Database—a brand new project from the Open Home Foundation—comes in.

Think of it as a source of truth created, curated, and validated by the community. A place where we gather everything from metadata (like power usage or infrared codes) to factual information (setup instructions such as how to add or reset a device) to real-world setup insights and community creations (such as automation examples). It will contain information users intentionally submit to the database, nothing will be collected automatically without your explicit consent.

It’s more than just having all device documentation in one place. We want to universalize all our collective experiences about devices in one place, which is more accessible, centralized, and structured than online websites and chat rooms.

With the Device Database, users both old and new can easily make well-informed choices and pick the best Home Assistant-compatible devices for them based on real usage experiences, allowing them to easily make decisions based on open home values — privacy, choice, and sustainability.

The data will be used and cross-referenced in Home Assistant and other Open Home Foundation projects, and acts as the backbone of their integrations. For example, a future infrared integration can hypothetically benefit from having access to readily available infrared codes from the database.

This isn’t something we can build alone. We’re relying on the strength of our global community to turn this vision into reality, and we’re backing it up with our infrastructure, engineering efforts, and partnerships:

  • The Open Home Foundation is here to provide the framework to help us collect and collaborate on knowledge easily and safely.
  • We’re working with Nabu Casa—our commercial partner—to create the most reliable hardware antennas that support more open protocols, such as Z-Wave, Zigbee, Bluetooth, and more, to expand the Home Assistant ecosystem.
  • The Works with Home Assistant program is being strengthened to make it easier to find trusted, high-quality devices for the platform, as you may have witnessed with the numerous new partners joining in the past few months.

The result? A system where no admin has to figure everything out in isolation on their own. Home Assistant can suggest, guide, adapt, and point users in the right direction, by drawing from the collective intelligence of the community (instead of scraping user data involuntarily like what Big Tech does).

And that’s not all for this year!

While device context and the Device Database are the major themes of this year’s roadmap, it doesn’t mean that we are dropping everything else that does not fit. We’re continuing work on other parts of the product experience:

  • A complete revamp of our automation triggers and conditions to make them both easier to use and more powerful.
  • A navigation and design system overhaul to improve feature discovery.
  • Continued improvements to make dashboards easier to use out of the box, including a new default dashboard and more.
  • Enhanced privacy controls for users, guests, and public access (Yes, I am aware of that.)
  • Easier setup for Music Assistant.
  • Make Assist more conversational, such as the ability to confirm and clarify a query.
  • Continued explorations in using LLMs to improve the overall user experience beyond voice.

In the meantime, we are working on making our roadmap more publicly accessible, so that you can discuss and track our progress with us! Stay tuned.

Laying the Groundwork for a Truly Smart Home

This roadmap isn’t about the flashy latest tech fad (what is it now anyway?) or abstract features. It’s about creating the enduring foundation to allow our community to build something greater than all of us together: a smart home platform that learns, grows, and adapts — with full respect for privacy, choice, and sustainability.

The central theme of the roadmap is that making Home Assistant smarter starts with understanding context, such as knowing what devices are and how they’re used, and we will use this collective intelligence to supercharge the main pillars of the user experience — automations, voice, and dashboards.

We’re making maintenance of a smart home easier, by making it less like a mandatory chore while still keeping the fun of tinkering (say, you wanna go wild on YAML), and by creating tools to help admins solve problems they don’t even know they have yet. We believe that, as our advanced users continue to tinker with their systems, their creations and discoveries will also benefit and elevate every user’s smart home.

We can’t wait to work on these initiatives for the rest of this year! Let’s build it together! 🚀👩🏻‍🚀👩🏼‍🚀👨🏼‍🚀

- Madelena, JLo, & Laura


This is a companion discussion topic for the original entry at https://www.home-assistant.io/blog/2025/05/09/roadmap-2025h1
17 Likes

I was wondering if energy management could make it onto the roadmap.

Many users use home assistant to optimize their energy for cost and sustainability. The energy dashboard is a very nice staring point to gain insight, bit it lacks control. Certain concepts like scheduling based on solar or dynamic energy prices would be a very welcome addition. Scheduling is relatively hard for beginners to get started with. I’m confident a lot of users would benefit from basic energy management functionality.

Maybe it would help if common devices like solar panels, home batteries and EV chargers would become standardize devices so it becomes easier to share blueprints.
Another topic frequently mentioned is to add power management. Power management is often overrated, but it is essential for load balancing and piek shaving.

17 Likes

Sounds promising. For devices to be fully in context it is important to be able to nest or combine devices, or link separate entities that do not currently belong to a device.

For instance, my fridge isn’t really smart. I have the temp sensors, door sensor and leak sensor, and a power plug, but they belong to different integrations. And some of the sensors I added myself, such as thresholds. I would not want to ‘dump’ all entities in one device (2 door sensors and two temp sensors have multiple entities), but they do belong together to the same fridge.

I’d want to have some control over grouping, as one door and temp belong to the freezer part and the others to the fridge. Power is for them combined.

5 Likes

shaq-shimmy

14 Likes

A device database will be welcomed by everybody. Your plans for Voice, however, are oddly non-specific. “Make Assist more conversational.”

Huh?

At the moment it’s the most black-boxy part of Home Assistant and it sounds as if you intend it to stay that way, which is a shame.

To give a specific example, I think a lot of people would appreciate the ability to write an automation like this:

alias: Voice automation
description: Sends message to external speaker on error
triggers:
  - trigger: state
    entity_id:
      - assist.intent_output
    to: error
    attribute: response_type
conditions: []
actions:
  - action: tts.picotts_say
    data:
      cache: false
      entity_id: media_player.living_room
      message: Sorry, I didn't understand that
mode: single

But we seem to be preoccupied with LLMs.

Edit: Another specific example: it isn’t possible to override built-in sentences. This makes it impossible to create a custom sentence beginning “What’s the…”

And now, of course, this “automation” will be picked up by AI. In a couple of weeks the “solution” will be popping up all over the place. :rofl:

3 Likes

Core redesign scares me quite a bit. Breaking changes, a lot of them? I really hope it’s for the better, right now it’s possible to add “homeless” entities to the devices, but only via GUI and not via yaml (which is a showstopper for me).

But what always wanted is an ability to create a virtual “device” and aggregate sensors into it. I even “faked” some these devices via MQTT discovery, but if it’s a built-in functionally, that’s is great.

Would be also interesting to see how this “database” will be built without a telemetry.

2 Likes

No, old automations will remain the same. Just a new style of triggers

3 Likes

I hope it’s not “device triggers” which are unusable anyway

As long as dashboards are being reworked, is there any chance that Lovelace will stop sending every entity state change to every single client that’s viewing or has viewed any dashboard, regardless of whether those entity states were relevant to the dashboard? This is a massive source of performance issues on older or underpowered tablets for those of us with larger numbers of devices. There’s no good reason to send data that isn’t going to be displayed, and if there’s an emphasis on “intelligence” here, it sure seems like a good time to reconsider that design decision.

7 Likes
1 Like

The energy dashboard needs a bit of tlc for sure. It has long-standing issues with a solar/battery setup and there doesn’t seem to be any traction on fixing them.

And yes, next logical step would be scheduling based on all of that data.

What long standing issues? Tim usually fixes many things in the energy tab.

This! I would love to add a few entities to one of my device that’s already in home assistant. Or make a new device with few different entities.

I would really like to see information on the timeline for implementing streaming voice response generation on this roadmap.

I agree with, and appreciate, all of this. I love where Home Assistant is going. What I do get frustrated with, from time to time, is how everything feels splintered at the moment. There’s HA Core for a lot of things, voice pipeline for TTS/STT, Music Assistant, and ties to LLMs.

If we want a true smart home, those things should be tied together closely in the background with transparent-to-the-user information sharing. Core and an LLM can monitor things, prompt users to take action if necessary by initiating a conversation via a voice pipeline/LLM, and respond to the user needing things like weather and music without a massive overhead of prompts and information passing every time there’s a LLM interaction.

That utopian future aside, what OHF, Nabu Casa, and the entire team is doing is great, the user base loves it, and we all applaud the hard work everyone is doing! Keep up the great work - we can’t wait to see what’s next!

2 Likes

100% in agreement with @AJediIAm

I have solar panels, a pair of Solis inverters, a pylontech battery stack in a Victron based system, and 1/2 hour octopus agile pricing.

The only reason I got into HA at all was to manage the charging schedule vs the Octopus pricing in winter.

It was a steep learning curve, and still isnt perfect.

I dont want to pay for electricity overnight if the forecast tomorrow is sunny and the solcast is for many hours (kWh!) of solar energy.

I’ve not personally got that good of an understanding that I can easily set up that scheduling for differences in weekday vs weekend consumption without taking time off work and dedicating a couole of full day weeks to learn how.

I like HA and Nabu Casa, and I built and use a lot more than just energy management, but energy is the cost saving driver that has paid for itself over and over again.

The reality today is that, according to our community survey, just 46% of partners and only 27% of children of long-time Home Assistant users are directly interacting with Home Assistant. That’s a big gap, and it highlights a broader challenge: Even the most experienced smart home administrators don’t always have the complete picture of their household’s needs.

Is that a valid conclusion?

The other people in my house don’t interact with HA directly but I believe I am meeting their needs. E.g.

  • when they walk into the room the lights come on and stay on until they walk out
  • when the washing machine finishes they get a reminder to empty it until it has been emptied
  • before they even get up in the morning the garden has been watered

My household would generally like our home to quietly get on with what it needs to do without having to interact with it.

When we do need to interact with it it is generally via Alexa… so even that is not “directly interacting with Home Assistant”.

14 Likes