Great 
. Thanks, have a nice evening too!
Hi Dimitri, I’ve just released version 2.10.0 of the app with two new trigger events (one for plain text prompt and one for text+image prompt). Let me know if everything works.
Hi Simone,
I tested it right away. It’s great that these trigger cards have been added. Here’s what I’m noticing:
If one camera is triggered and receives the response with the analyzed image, it works well. I get the correct response along with the correct analyzed image.
If two cameras are triggered at the same time (which happens regularly), Gemini does return the correct responses — so the text for camera 1 and camera 2 is accurate. However, the images are the same for both cameras. In that case, both camera 1 and camera 2 receive the analyzed image from camera 1.
Other than that, it works perfectly!
You only have this problem with the new trigger I just added, right? If you use the image directly from the action card, does everything work correctly?
Yes that’s correct.
This is a compromise I had to make to prevent every analyzed image from being saved to Homey, which would cause the app to crash at some point. Now the app only saves one image at a time (which means that in the case of a simultaneous call, the second image overwrites the first). I just uploaded v2.10.1, where I added in Settings->Configuration the option to specify the number of simultaneous images the app can handle (the default is 4, so Homey always has four preloaded images). Be careful, because the higher the number, the more Homey memory the app uses.
Let me know if it now works with simultaneous calls (I don’t have a camera and haven’t been able to do extensive testing).
Version v2.10.1 works perfectly. I tested it with 4+ cameras triggering at the same time. I now receive the correct image with the correct text and camera. Thank you!
Does this mean that all images are stored in memory, or are they deleted after a certain period of time? To prevent the app from crashing after a few weeks/months due to too many images in memory like you said.
Thank you very much for the improvements, Simone — the donation is on the way!
Hi Dimitri, thanks for the donation!
Now, by default, the app saves 4 image slots for Gemini in Homey. You can set a higher or lower number in the app settings. If you have 16 cameras in your home and set this value to 16, you’ll never have problems even if a presence trigger were to trigger for all 16 at the same time. However, this does mean more memory is used (how much depends on the photo resolution). If everything is working perfectly now, I suggest leaving the default value (4 slots). If you encounter any issues, you can always increase it.
If you want to know which and how many images Homey is saving to memory, you can check https://tools.developer.homey.app/ (images section).
Have you also tried the action card for managing your smart home? If you use it together with the Telegram Homey app or WhatsApp Homey app, you can actually chat with your home and ask for information or send commands.
Hi Simone,
Clear, I can indeed see the images in the image section.
Good tip — I’ll test that next once all the cameras are set up .
Can I also create a summary in Telegram using Gemini and/or automatically tell the weather every morning?
Hi @Mark1541 , this app doesn’t have internet access, so it can’t tell you the weather or the main news of the day. It can manage your devices in Homey and provide reports to keep you updated on the situation at home (what’s on/off, if any doors are open, how much temperature it is, etc.)
Just a brainfart: maybe it can use the available tags of any installed weather app @ Homey to create a nice weather forecast message?
You could actually use Homey’s “date and time” trigger associated with a weather app to get the weather every day. The forecast tag could be passed to Gemini with a message like “tell me today’s forecast in a fun way [weather tag].”
I like the idea of this, can you give an example how id do it with a weather app?
Thanks
Very cool app. Any plans for an app that can use local LLMs like Ollama?
Hi, I don’t plan to create apps for other LLMs, but I see that ollama is already supported by other developers on the store: App Store di Homey
I can’t help you with details because I don’t use weather apps.
Either collect all the tags from the weather app, pass them to gemini ai (trigger flow “send the prompt…”), and in the prompt ask it to create a weather forecast based on this information. Or, if the weather app doesn’t just provide data but also provides a conversational forecast, you can pass that directly to Gemini, asking it to rewrite it in a friendly tone or something similar.
Thank you for your response and the work you have done for the community.
Hi @s_dimaio,
I received an email from Google stating that the preview version of the Gemini model we are currently using will soon be discontinued.
The following change is required:
-
Old version: gemini-3.1-flash-lite-preview
-
New version: gemini-3.1-flash-lite
According to Google, the preview version will stop working on May 25, 2026. After that date, API requests to the old model will fail. Google also mentioned that the new GA version uses the same underlying architecture, so no changes to prompts or application logic are required.
Could you please add the new model version and update the current configuration to:
gemini-3.1-flash-lite
Thank you in advance.
Kind regards,
Dimitri
