In the Netherlands, Energy prices tend to fluctuate sharply during winter often to extreme 0.5 per kwh. Maintaining control over them requires access to timely and structured data. My energy provider have their own app, but it is limited to the mobile phone. My goal is to set up a system that allows direct querying of energy price information so I can plan and optimize energy usage more effectively especially when planning LLM.
I fetch prices, store them DuckDB for local analytical storage. It runs automatically in the afternoon when tomorrow prices get updated. This combination allows data to be pulled, analyzed, and visualized in a simple Streamlit for a simple dashboard interface. It has a simple rule to check for average and flag hours that don’t are higher than mean.
The next step is to connect these data pipelines to an MCP server that will provide access to an LLM. The model will use the data to generate a daily report and schedule recommendations based on current and predicted energy prices. I also want to integrate weather forecast into the whole thing and see if I can implement a model or two that can predict prices.