← Back to context

Comment by dzonga

3 days ago

what strikes me about these posts is they praise models for apps | utilities commonly found on GitHub.

ie well known paths based on training data.

what's never posted is someone building something that solves a real problem in the real world - that deals with messy data | interfaces.

I like a.i to do the common routine tasks that I don't like to do like apply tailwind styles but being renter and faking productivity that's not it

I used it with gemini 3 in tandem to build an app to simulate thermal bridges because I want to insulate a house. I explored this in various directions and there are some functionalities not completed or sound, but the main part is good and tested against ISO/DIN test cases for this kind of problem. You can try it here, although the numeric simulations take quite a while in the cloud app

https://thermal-bridge.streamlit.app/

Disclaimer: I'm not a programmer or software engineer. I have a background in physics and understand some scripting in python and basic git. The code is messy at the moment because I explored/am still exploring to port it to another framework/language