Home Assistant Users on a Time-of-Use Rate
If you already run Home Assistant with a thermostat, EV charger, water heater or battery connected, the integration that fits is our HACS package: it reads the climate, switch and sensor entities you already have, builds a 24-hour schedule each night against your utility's rate, tomorrow's forecast and a thermal model trained on your own readings, then hands the plan back to your own automations to run. There is nothing to replace and nothing new to buy. The free tier runs the full optimizer on one HVAC device; Premium removes the device limit and adds the rest.
Who this is for
- Home Assistant users with a
climate.*entity for a thermostat they already control from HA - Home Assistant users with an EV charger, home battery or water heater exposed as a switch entity
- Households on a published time-of-use or dynamic rate, PG&E, SDG&E, SCE, ComEd, Ameren and more, who want the daily scheduling handled instead of hand-written automations
- Households with solar who want a production sensor's real output feeding the forecast, not a nameplate estimate
Not for you yet if you have no climate or switch entity to connect, or if your only unit is a window AC or mini-split with nothing but an infrared remote, that's the Smart Satellite, a separate product with no Home Assistant required.
What Hungry Machines does in this situation
Install through HACS, sign in with your hungrymachines.io account, then open the panel and add each appliance: map the HVAC's climate entity (and optional indoor-temperature, humidity and power sensors), the charger, battery or water heater switch, and pick a weather entity. Choose your pricing, a preset time-of-use plan or your own hourly numbers, then set comfort and charge limits per appliance: a base temperature and savings level for HVAC, a target charge and deadline for the charger or battery, a temperature range for the water heater.
Your instance pushes a sensor reading every five minutes and pulls back an optimized schedule once a day, applied on half-hour boundaries; in between, a five-minute check nudges the setpoint a degree at a time if a room drifts more than half a degree off its target.
The panel shows each appliance's schedule, today's estimated savings, and a diagnostics view that flags a stale sensor instead of quietly optimizing on bad data; the hm-thermostat-card and hm-savings-card Lovelace cards put the same information on your existing dashboards. Override any slot, or pause one appliance's optimization without touching the rest.
The arithmetic
- Heating and cooling run about half of a typical US home's electricity use, which is why HVAC is the largest single lever a TOU rate offers.
- A tuned pre-cool-and-drift schedule typically cuts HVAC energy cost 15-30% on a TOU rate, roughly 8-15% of the total electric bill.
- A thermal model built from your own home's readings measured up to 80% more accurate than a pooled, generic-house model in our own analysis.
- In pilot data, a fitted model reached 45-77% better next-day accuracy than assuming temperature doesn't move by day two or three of connecting; on days with room to optimize, projected HVAC savings ran 25-60% (median around 40-45%), with roughly 3 days in 10 shipping 0% honestly.
These are the modeled and pilot-fleet figures described in our HVAC-automation and onboarding guides, not a guarantee for your specific home, your own result depends on your rate's spread, your comfort band, and how much thermal mass your house holds.
How to set it up
Add https://github.com/hungrymachines/energy-dashboard as a custom repository in HACS, download, and restart Home Assistant; then add the integration from Settings → Devices & Services and sign in. From there, the panel walks you through registering appliances, see the full Home Assistant integration page for the complete install steps and what Free vs. Premium covers. Prefer not to run this inside your own Home Assistant instance? The Home Hub runs the same optimizer on its own hardware instead. Either way, check pricing for the current plan details.
Read the full guides
- Home Assistant integration, the full product page: install steps, what it reads, Free vs. Premium
- What Happens After You Connect: Your First Two Weeks
- Automating Your HVAC for Time-of-Use Savings
- Building your own client instead? Build a client with an AI agent walks through the public API.
Frequently asked questions
Which smart thermostats and devices can the integration control?
Any device already exposed as a Home Assistant entity. For HVAC, that's most modern smart thermostats, ecobee, Honeywell, Nest with some limitations, and any Z-Wave, Zigbee or Wi-Fi thermostat Home Assistant already supports. Beyond HVAC, the panel also registers an EV charger, home battery or water heater as a switch entity, plus solar as a production sensor with nothing to switch. Each HVAC gets mapped to its own climate entity and its own thermal model, so a second unit in the house is a second appliance card, not a shared setting. If a device has no Home Assistant entity at all, a window AC with only an infrared remote, for instance, that's the Smart Satellite's job, not this integration's.
How soon after connecting does the schedule actually save money?
Readings start flowing the moment you connect, and nothing about your home changes on day one, that's purely data collection, plus one optional guided calibration window on the first suitable day. By day two or three, the fitted thermal model has typically become 45-77% more accurate than assuming the temperature simply won't move, which is the bar it has to clear before a plan ships at all; that's also when the first optimized schedule appears. Projected HVAC savings on days with room to optimize typically run 25-60% against a fixed setpoint, though roughly three days in ten ship 0% honestly, when prices are flat or the day leaves nothing worth shifting. The model keeps refitting nightly through the first two weeks, then settles to weekly.
Does this work the same way for heat pumps as for central AC?
Yes, and heat pumps tend to benefit more, not less. A heat pump's efficiency varies with outdoor temperature, so running it during a warmer off-peak hour, when its coefficient of performance is higher, saves more energy than running the same cycle during a cold on-peak hour would. The per-home thermal model the integration builds from your own sensor readings accounts for that automatically; it doesn't assume your equipment behaves like a generic furnace or a generic compressor. The schedule still respects whatever comfort band you set, and the weekly refit keeps the model current as the season, and the heat pump's typical operating point, shifts.
About Hungry Machines
Outsmart your electric bill - Hungry Machines Energy provides an energy management system that shifts HVAC, EV charging, home batteries, and other electric loads to cheaper hours for homes and buildings with solar power or variable electricity rates.