Home Assistant follows a monthly release cadence, typically where the first Wednesday of every month brings a new update, and the last Wednesday of every month is the beta for that release. October's update is now here and rolling out to every user.

I typically install the beta every month (after waiting a day or two for the initial fixes and bug reports), so I've already been using it for the past week. These are the most interesting features, and the ones that may revamp your Home Assistant setup.

Automations are getting further improvements

Resizable sidebar and overflow options moved back

Home Assistant's September update brought a rather big change to automations: the introduction of a sidebar. When clicking an action in your automation, it now opens a sidepanel on the right-hand side for easier inspection and modification, making use of all of the unused screen space that previously made up most of the display in the automation editor.

At the time, in order to delete an action, you had to click the block, then on the right-hand side, click the delete button inside the block editor. The overflow menu for blocks only contained "move up" and "move down" options, but you could still drag blocks around if you prefered. That's changing now, and the control options have been moved back to the individual blocks, reducing the number of clicks.

However, that's not all. The sidebar is now resizeable in order to give you more room for YAML-heavy editing, and you can now copy and paste blocks with CTRL+V if you want to reuse a block. Furthermore, there's now undo and redo support, invoked via CTRL+Z and CTRL+Y. All of these will make creating and editing automations significantly easier.

Adaptable dashboards learn your usage

Suggested entities

Home Assistant is getting an all-new adaptable dashboard packed with suggested entity actions. I spotted this initially as a part of the "usage prediction" tree on GitHub, and I was curious to see how exactly it would manifest. As it turns out, it's a part of the "Home" dashboard, which generates a set of controls per area, and surfaces your most commonly used controls.

These controls are surfaced based on how often you use them, and calculates whether to show them based on the hour of the day. For those who don't want to take the time to build their own dashboard, the Home dashboard is a great way to get something up and running quickly that works.

If you want to add these predictors to a normal dashboard, you can edit the YAML and add the following to a section:

strategy:
 type: common-controls
 title: Common controls

I was unable to get this working on the final beta version, but it should be in the stable release.

Voice assistants get a second wake-word

Two separate assistant pipelines

For a long time, ESPHome-based voice assistants (even tiny satellites like the Atom Echo) have quietly supported multiple wake words in the background. With this release, that feature is finally available for everyone to configure.

You can now set up two wake words and two assistants for each device in your home straight from Home Assistant. That means a dual-language household can assign different wake words to different languages, like seting " Okay Nabu" for French and "Hey Jarvis" for English. Alternatively, you could keep a separation between your local and cloud-based assistants: "Okay Nabu" could route to a cloud pipeline, while "Hey Jarvis" stays local.

It's a flexible system, and one that opens up plenty of possibilities for tailoring how you use voice around the house. On top of that, commands that are relevant to the room you're in will respond with a beep when they've been completed, rather than a "Turned on the lights" or "Brightness set." It's a quick confirmation, as you probably have visual confirmation that the task was completed in front of you anyway.

The only limitation is that this doesn't extend to AI-enabled Assistants, since they generate dynamic responses that can't be replaced with a single tone.

Built-in support for AI image generation

With this update, you'll also be able to include AI image generation pipelines as a part of your automations. Whether it's entirely newly generated images or processing to your currently-existing ones, you can include it as a part of an automation and serve up the response.

An official example of the integration in action makes use of a doorbell snapshot to turn the camera output into a cartoon image, sending it to the user's phone. There are a lot of creative ways to use this, and it could be a great way to dynamically generate images for weather notifications or displays powered by ESPHome.

New and improved integrations

New smart devices and a Portainer integration

As always, the community has been busy adding support for even more devices and services. October's update brings a solid batch of new integrations, ranging from smart lighting and water monitoring to biometric access systems and Docker management.

Highlights include:

• Cync for connecting GE smart lighting devices.
• Droplet to track real-time household water usage.
• Libre Hardware Monitor for keeping tabs on PC hardware sensors.
• Portainer to monitor and manage Docker containers directly.
• SFTP storage for remote backups.
• Victron Remote Monitoring to pull in solar and energy data from VRM.

On top of that, there are a bunch of improvements to existing integrations, like AccuWeather, Reolink, ntfy, Shelly, SwitchBot, Tasmota, and Tuya. All in all, it's a pretty sizeable update with a lot to love and with something for everyone.