About EcoHome

EcoHome is an AI agent that helps you run a smart home more cheaply and with a lower carbon footprint. Ask it when to charge your EV, run an appliance, or make the most of solar, and it gives a specific, time-based recommendation backed by live data.

What it does today

The advisor reasons over two live data sources for your selected UK region: half-hourly electricity prices from the Octopus Agile tariff, and hourly solar irradiance from a weather forecast. From these it works out the cheapest and greenest hours and turns them into plain-language advice, for example the best window to charge an EV tomorrow to line up with low prices and peak sun.

It runs as an LLM agent with a small set of tools: live pricing, a solar and weather forecast, a knowledge base of energy-saving best practice, and a savings calculator. It picks the right tools for each question, then synthesises a recommendation with the numbers to back it up.

This is a demonstration. It runs on a sample household dataset, not real smart-meter readings, so its usage figures are illustrative. Prices and solar forecasts, however, are live and region-accurate.

Where it's heading

Roadmap

These are planned and exploratory directions, not current features.

  • Real smart-meter integration. Connecting to a household's actual consumption data (for example via the Octopus consumption API), with the user's consent, so advice can be grounded in how the home really uses energy rather than a sample profile.
  • Personalised usage insights. Once real usage is available, spotting a household's own patterns and suggesting specific, measurable ways to shift load to cheaper, greener hours.
  • Automated scheduling. Moving from advice to action, letting the agent schedule compatible devices (EV chargers, batteries, appliances) directly around price and solar forecasts.
  • Wider tariff and region coverage. Supporting more energy tariffs and markets beyond UK Agile pricing.

Research & code

This work is described in an accompanying paper:

S. Jonah (2026). LLMs for Agentic Home Energy Management. arXiv preprint arXiv:2607.04569.

Preprint — the peer-reviewed version will replace this link once published.

View the source on GitHub

EcoHome is a portfolio and research demonstration. Its recommendations use sample data and are not a substitute for professional energy advice.