cheap-watts open-sources AI-assisted home energy control
Developer Romin Irani open-sourced cheap-watts on GitHub, using Google Antigravity with Gemini 4 Hydrogen and Gemini 4 Carbon to coordinate rooftop solar, a home battery, two vehicles, and dynamic electricity tariffs. The project includes offline tests across more than 25 commits, hardware abstractions, CI safety checks, and a dry-run mode that queries live inputs without writing to the inverter. Irani reports that the models handled requirements, coding, and testing while he focused on architecture and reviewing changes.
The account provides a concrete, open-source example of using multiple AI models and safety gates to automate a physically consequential home-energy workflow.
Running rooftop solar, a home battery, two electric vehicles (EVs), and dynamic hourly electricity tariffs requires tight coordination. Without automation, these systems work against each other.
At home, I run dual east/west solar panels feeding a hybrid inverter with a 10 kWh battery, alongside a plug-in hybrid, an EV, and dynamic tariffs from my electricity provider. On sunny days, you should never buy grid power when solar can charge the battery for free. On dark, cloudy days, you want to buy grid power only during the cheapest hourly windows. Moreover, if you charge the home battery from the grid before plugging in an EV, the car drains your charged home battery right away.
Last weekend, in a couple of hours of hands-on work starting Saturday evening, I engineered cheap-watts to automate this workflow and open-sourced it on GitHub, complete with a full offline unit test suite across more than 25 commits.
The best part? I almost never touched the code myself.
By pairing Gemini 4 Hydrogen and Gemini 4 Carbon inside Google Antigravity, I practiced agentic engineering and stayed focused on high-level architecture. Here are the three main takeaways from the weekend build:
#cheap-watts#Gemini 4 Hydrogen#Gemini 4 Carbon#Google Antigravity#GitHub