The dashboard opens on the Overview — the whole fleet in a glance: counts per heat level, coverage gaps (unassigned sensors, sensor-less stores), the last seven days of alerts, and one card per region, worst first. Every number is a link into the matching detail view. The Grid, Stores and Map views below it exist because five units and five hundred units are genuinely different problems.
- Grid — a card per unit with its current reading. The right choice up to a few dozen units.
- Stores — units gathered under the store they protect, which is how most people think about a restaurant estate.
- Map — geographic, useful for spotting that a whole city is having a bad afternoon.
- Regions — rolled up by region, for when your span is too wide to read store by store.
- Triage — only what needs attention right now, worst first.
Whatever the view, a unit that has crossed a threshold sorts toward the top. You should never have to hunt for the bad news.
Above the views sits a 7-day forecast banner, built from Open-Meteo weather and each unit's own history. When it predicts trouble it lists the units and time windows; when everything is clear it says so — and the all-clear can be snoozed until tomorrow if you'd rather not look at good news all day. A forecast that predicts heat problems always shows, snoozed or not.