Ciao @DimitriEbben, the updated model is now being tested along with other improvements!
Simone
Ciao @s_dimaio,
Thank you for the fast update!
I have tested the new version and it works good, also with the automatic conversion from the preview to the GA.
Thanks a lot!
A new version of the app is currently being tested. In this version, I’ve added a widget for chatting with Homey from the app’s home screen:
Hi @s_dimaio , thank you for this great app 
! I’m trying to use Gemini AI for my sunscreens to analyse the weather. But i have learned that the API has standard no internet access to look up the weather data. Is it Possible for you to activate “googleSearch” in the API call?
Or maybe an option to activate that in the app settings?
Ciao @Lars_Faber , after introducing the widget for chatting with Gemini, I’m now implementing flow management. I’m testing the ability to create new flows (only standard ones for now) simply by requesting the desired automation in the widget.
I’ll implement Google search immediately afterward and add the new 3.5 Lite model! Both should be relatively fast.
@s_dimaio Thanks for your fast reply! I’m exited that you are cooking the app so fast . Nice work.
Why is it possible with the open ai ChatGPT app?
Mainly because it’s a different app. However, as mentioned, supporting Google Search was already on the roadmap, and I plan to release a first version to the test channel this evening.
Ah Okay.
Hi everyone, these are the latest new features I’ve added to the app:
-
stable version 3.0.0: I’ve introduced the ability to manage Flows (currently only the standard ones) directly from Gemini. Using the chat widget or the smart home flow, you can ask Gemini to create any automation. You can also ask to modify or delete existing ones (in these cases, Gemini will ask for confirmation before proceeding);
-
test version 3.1.0: I’ve added access to Google search for all Flows so you can get up-to-date information. I’ve also updated the Gemini models and added Gemini 3.5 Lite.
Thank you will test it
Thanks I installed the test version!
Is it true that I don’t get any options without advance flows?
I tried to test it with telegram but it remains empty The options at Gemini?
I’m not sure I understand. Gemini action cards only support advanced flows because they return a token that can be used by subsequent flows. Action cards in standard flows don’t support tokens.
The new access to google search is great. Made a few flows telling me weather and traffic info for my commute and works great. Trying to think of other things I can use this for.
Great job with the latest update to the app.
I’m curious in general what they’re doing to use Gemini for!!!
With the test version i get the 429 error “To many requests”.
One request did 19 API calls, what i can see in the dashboard.
You have exceeded your daily quota. Please try again later or upgrade your plan. 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.
I used the Gemini 3 flash , and also try to switch to the 3.1 flash lite.
Only one flow active with at 10:00AM a weather check.
Changed it back to the 2.5 lite and seems to work.
Is this burning that many tokens?
Edit: I think i see why, the higher models don’t have search grounding
Can you share the flow ?
Sorry but the text is in Dutch.
But i start at 10:00AM and then ask gemini:
Analyze the weather forecast for Doetinchem for the rest of today. Does it meet these three conditions: the expected high temperature is > 20°C, the chance of precipitation is < 15%, and the wind speed is low enough to safely open the sunscreen? Answer only with “true” or “false” and do not provide any further explanation.
Then i convert the reaction to a yes/no variable.
If the result is yes, then i do an additional check for windspeed and Rain. With the Openweather and Buienalarm app.
If the wind goes beyond 15 km/h or rain is detected within the hour or raining then put the sunscreen back in.
