Skip to content
Hungry Machines Docs

Energy Dashboard for Home Assistant

This is the recommended path for most users. The Energy Dashboard installs through HACS, signs in with your Hungry Machines account, and handles sensor collection plus schedule application automatically. No code, no API keys to manage.

If you want to extend the default dashboard with your own cards or automations, see Custom Home Assistant integration. If you don’t use Home Assistant at all, see Build Your Own Client.

What you need

Section titled “What you need”
  • Home Assistant (any installation type — HAOS, Container, Supervised, Core)
  • HACS installed and configured
  • Temperature sensors and an HVAC climate entity (the optimizer’s minimum input set)

Step 1 — Create your Hungry Machines account

Section titled “Step 1 — Create your Hungry Machines account”

Go to hungrymachines.io/signup and sign up. Confirm your email when the verification message arrives. The email and password you set here are what you’ll use to sign in inside Home Assistant.

Step 2 — Install the Energy Dashboard via HACS

Section titled “Step 2 — Install the Energy Dashboard via HACS”

  1. Open Home Assistant → HACS in the sidebar.

  2. Click the ⋮ menu (top-right) → Custom repositories.

  3. Add the repository:

    • Repository: https://github.com/hungrymachines/energy-dashboard
    • Type: Dashboard
    • Click Add.

  4. Find Hungry Machines Energy Dashboard in the HACS list, click into it, and click Download.

  5. Restart Home Assistant when prompted.

Step 3 — Add the dashboard to Home Assistant

Section titled “Step 3 — Add the dashboard to Home Assistant”

  1. Settings → Dashboards → Add Dashboard → From scratch (or your preferred method).

  2. Pick the Hungry Machines Energy Dashboard template that now appears in the list.

  3. Save.

The dashboard shows up in the sidebar.

Step 4 — Sign in

Section titled “Step 4 — Sign in”

Open the dashboard. The first time you visit, it’ll ask for credentials — use the same email and password you set up at hungrymachines.io.

Once signed in, the dashboard:

  • Pushes sensor readings to the Hungry Machines API every 5 minutes
  • Pulls a fresh optimization schedule each morning
  • Applies the schedule to your thermostat (or whatever climate device you’ve connected)
  • Shows your current vs optimized usage, savings estimate, and per-interval setpoints

What to expect

Section titled “What to expect”
DayWhat happens
Day 1Dashboard starts collecting readings. Schedule shows safe defaults based on your preferences.
Days 2–3Thermal model accumulates data. Defaults stay in effect — comfort is maintained, no savings yet.
Day 4 (typically)The daily initial-fit job (06:30 UTC) builds your first per-user thermal model as soon as you have ~72 hourly buckets of data. That night’s optimization run produces your first source: "optimization" schedule with non-zero estimated_savings_pct.
Week 1+Sunday 02:00 UTC weekly refit keeps the model current. Savings ramp up as the model sees more weather conditions and HVAC cycles.

The three-day warm-up is normal. The optimizer learns each home’s specific thermal behavior — pooled or generic models are 14–82% less accurate per our analysis — so the first three days are about gathering enough data for the fit to be meaningful. Gaps in readings (sensor downtime, network outages) push that timeline out.

Updating the dashboard

Section titled “Updating the dashboard”

When a new release ships, HACS shows an update badge. Click → Update → Restart Home Assistant. Your sign-in and preferences persist across updates.

Troubleshooting

Section titled “Troubleshooting”

Sign-in says “invalid credentials”

Section titled “Sign-in says “invalid credentials””

Try signing in at hungrymachines.io/login first — if that works, the credentials are correct, and the issue is between HA and the API. Check that your HA instance can reach https://api.hungrymachines.io (corporate firewalls / split DNS are common culprits).

If the website login also fails, reset your password.

”Email not confirmed”

Section titled “”Email not confirmed””

Check spam. Click the verification link before signing in.

Schedule still shows defaults after a week

Section titled “Schedule still shows defaults after a week”

Gaps in readings (network drops, sensor downtime) slow training — the fitter needs at least 72 hourly buckets of data within the last 14 days to produce a model. See Troubleshooting for the full diagnostic.

Other issues

Section titled “Other issues”

File a GitHub issue on the dashboard repo, or email info@hungrymachines.io.

Next steps

Section titled “Next steps”