[APP][Pro] Gemini AI - Control your Smart Home with Google's Gemini

Hi everyone,

I’m excited to share Gemini AI, an app that brings Google’s latest generative AI models directly into your Homey Pro. This app allows you to move beyond simple automations and interact with your smart home using natural language, image analysis, and intelligent reasoning.

:house: Natural Language Smart Home Control

The core of the app is the “Send a command to your smart home” Flow card. Instead of building complex logic for every scenario, you can simply ask Gemini to manage your home for you.

Basic Device Control/Inquiry

  • Device Control: “Turn off all lights except the kitchen,” “Set the thermostat to 21 degrees,” or “Open the garage door.”

  • Status Queries: “Are any windows open?”, “Is the living room TV on?”, or “What is the temperature in the bedroom?”

  • Multiple Actions: “Turn off the lights in the bathroom and start the coffee machine.”

Advanced Device Control

Beyond basic controls (on/off, brightness, temperature), Gemini can now access all Action Cards for every device:

  • Full Automation: Every function of every device is now controllable through natural language.

  • Device-Specific Commands: “Reset the energy meter,” “Start the delicate wash program,” “Set the air conditioner to dehumidify mode,” “Activate night mode on the camera.”

  • Automatic Discovery: Gemini automatically discovers all available Action Cards for each device and executes the appropriate one based on your request.

[NOTE] Action Card controls require HomeyScript installed from the Homey App Store.

Flow Management

Gemini can discover and trigger your Homey automations:

  • Flow Discovery: “What flows do I have?” or “Show me my automations.”

  • Flow Triggering: “Start the Good Morning flow” or “Activate Movie Time.”

  • Advanced Flows Support: Works with both standard and Advanced Flows (Homey Pro).

  • Flow Details: “Tell me about the Evening Lights flow.”

[NOTE] Flow Triggering require HomeyScript installed from the Homey App Store.

Scheduling & Timers

Gemini can handle time-based commands natively. You can send commands like:

  • “Turn on the heating tomorrow at 7:00 AM.”

  • “Turn off the garden lights in 20 minutes.”

  • Management & Precision: You can view and manage all active timers in the App Settings.

[NOTE] For timers exceeding 24 hours, please allow for a small execution tolerance of up to 10 minutes.

:camera_with_flash: Multimodal Intelligence (Image Analysis)

Use the “Send Prompt with Image” card to let Gemini “see” what’s happening through your cameras.

  • Monitoring & Security: “Is there a person carrying a package in this snapshot?” or “Is the iron still on in this picture?”

:brain: Pure Logic & Creativity

Of course, with the “Send the Prompt” card you can still use Gemini for text-only processing in your flows:

  • “Generate a welcoming message for my guests.”

:money_bag: Costs & API Keys (Free vs Paid)

To use the app, you need an API Key from Google AI Studio.

  • Free Tier: Google provides a generous free quota, but its “real-world” capacity depends on how you use the app:

    • Text & Image Prompts: You have plenty of room for daily text or image analysis.

    • Smart Home Commands: Since “Function Calling” requires multiple API exchanges to understand your home’s context, the free tier typically allows for 5 to 10 actual commands per day, depending on the complexity of the request.

  • Paid Account: For daily, unrestricted use, a Billing Account is highly recommended. The Gemini 3.1 Flash-Lite model is extremely cost-effective.


:locked: Privacy & Permissions

  • To manage and query your devices locally the app requires the permission:

    homey:manager:api
    
  • For triggering Flows and executing advanced device Action Cards, you must install the HomeyScript app from the Homey App Store.

  • Prompts and images are sent directly to Google’s API; no data is stored or logged by the app.


:clipboard: Flow Examples

Here are some practical examples to get you started:

Smart Security Monitoring

WhatsApp / Telegram Integration

“Ask a Question” Notification

Alexa Integration


Download it now on the Homey App Store! Feel free to share your feedback, interesting use cases, or any questions you might have.

Simone

Test version: Gemini AI | Test

2 Likes

Great app, Been using this with my Reolink Doorbell and describes who’s at the door via a push notification.

Works very well. Was using similar integration in HA which is what I missed about HA but now can be done in homey. Thanks for making this

Thanks! Have you had a chance to try out the latest test version? I’m trying to figure out whether it’s worth continuing to develop a flow that Gemini can use to manage all the devices in Homey. I’m interested in seeing if there are any real-world uses for this flow.

I have , It works perfectly.

How limmited is the free tier?

I have an subscription on th eGemini pro!

Hi Mark, Gemini Pro (which I also have) is not connected to Gemini API Keys.
The free API Key plan allows approximately 5-10 calls per day for home automation management (depending on the complexity of the requests).

You can still test it with free plan: when you reach the daily limit, Flow will return the error “You exceeded your current quota, please check your plan and billing details. For more information on this error, head to: https://ai.google.dev/gemini-api/docs/rate-limits. To monitor your current usage, head to: https://ai.dev/rate-limit.”

Do you get names as an answer or just a description? Im looking for an alert with names. Is there any way to train gemini with this?

Just a quick description. As far as I know, face recognition isn’t supported in the public Gemini API yet.

Hi everyone,

I’ve just released version 2.1.0 of the app! This update introduces a major new feature: you can now schedule and manage your home automation commands directly through Gemini AI.

You can use flows to ask Gemini to handle specific tasks for you. For example:

  • “Turn on the heating tomorrow at 6:00 PM”

  • “Turn the lights off at 11:00 PM”

Give it a try and let me know how it works for your setup!

Hi everyone,

Just a quick heads-up: Version 2.2.1 is now in test! You now have full control over which Gemini model you want to use. You can now toggle between:

  • Gemini 2.5: Flash-Lite, Flash, and Pro
  • Gemini 3 (Preview): Flash and Pro

Check it out and let me know if you have any feedback!

Simone

Gemini AI for Homey just got smarter with version 2.3.0 now in test.

The main highlight is Conversation Memory: Gemini now remembers the context for up to 2 hours. This is perfect for a combined use of Gemini AI with one of the Homey messaging apps, allowing you to have natural, multi-step conversations without repeating details.

This is an example of a flow for a joint use of Gemini AI and Homey Telegram app to control your smart home::

Changelog complete v2.3.0:

  • Conversation Memory: Remembers context across multiple interactions (2-hour history).
  • Recursive Zones: Automatically finds devices in sub-areas/sub-zones.
  • Virtual Class support: Correctly identifies sockets/switches configured as lights.

Simone

v2.4.1 Stable is here!

  • Flows: Now supports Action Cards for every device via HomeyScript;
  • Smart Home: 11 new capabilities for better device search and zone control;
  • Speed: Improved performance, timezone accuracy, and API limit management.

Great App! BTW: What Alexa App did you use in your Alexa integration example? It seems to be able to get the spoken text back as text/json?

Hi!
This is a project I worked on a while ago when the first Alexa app for Homey seemed to be abandoned, but I never ended up publishing it.
The app is actually able to intercept both the command sent to Alexa and Alexa’s response. If you’re interested, you can find it here (you’ll need to download the code and do a manual installation):

However, I have to admit the Gemini integration isn’t perfect: while everything functions exactly as it should, Alexa automatically raises the volume on Echo devices when replying—even if they’re set to zero. Because of this, you’ll hear a snippet of Alexa’s ‘official’ response (‘I’m sorry, I can’t find that device’) right before Gemini’s ‘correct’ answer kicks in.

The integration with the Telegram app is definitely better!

Thanks! I had a brief look at it. It seems you’re using the conversation history to grab the text, therefore the delay. Appreciate there is no alternative way available other than maybe implementing an Alexa skill. Since the overall speech-to-text quality of Alexa devices is always better than any homegrown solution, this caught my eye :slight_smile: