We respect your privacy
Mneme-AI only uses cookies strictly necessary to operate the platform. You can customise your preferences at any time. Learn more
The specific problem of multi-site managers
Managing a single building is manageable with a flat dashboard. Managing 5 or 50 buildings is another story. You want to see a global health state at 9 AM, identify the 2 or 3 buildings worth your attention today, drill-down on the one blinking, and resolve priority alerts without manually navigating between 50 tabs.
Legacy BMS often require one instance per building, making portfolio view impractical. Sensia is natively multi-site in the Hypervision range.
The hierarchical Supervision canvas
An IoT workspace's analytics revolves around a hierarchical drill-down canvas. For smart buildings, the hierarchy is site → building → floor → room → equipment → point.
You start at the root level (all sites), click on a site to enter it, click on a building to enter it, and so on. The breadcrumb at the top lets you go up one level with one click. The URL is the source of truth: ?path=site:UUID/building:UUID/floor:UUID — shareable.
Auto-computed KPIs at each level
At each level of the hierarchy, Sensia automatically computes the relevant KPIs:
- Total sensor count in the sub-tree
- % online (sensors that reported in the last N minutes)
- Active alerts filtered by severity
- Entities by slug (how many Buildings in this Site, how many Floors in this Building…)
KPIs are computed in real time via DigitalTwinService — no nightly batch, no stale data.
Auto-density drill-down
The direct children list at each level adapts density automatically:
- Cards (≤24 children) for a large-format view with preview
- Compact table (>24) to quickly scan a list of 100 buildings
- Search bar beyond 12 items, by name and all text/email/select attributes
- « Load more » pagination beyond 100
Clicking on a row drills down one level.
Real-time via WebSocket on every screen
Sensia pushes changes in real time via WebSocket — no 30-second polling. When an alert fires on a sensor, the UI updates sub-second on every connected screen.
The mechanism rests on the twin.{thing_id}.changed firehose pushed via Redis pub/sub to WebSocket pods. Typical latency: 100 to 300 ms between DB change and UI. On the frontend, smart debounce (100ms per parent_id) avoids wild re-renders on bursts.
Fine RBAC for multi-site facility teams
Want your Lyon site manager to only see Lyon buildings, not Marseille? Space-scoped RBAC: each space (site) has its own permissions (registry.view, registry.manage, alerts.acknowledge).
You assign users to relevant spaces. API requests are filtered via FORCE Row Level Security in Postgres — a Lyon user PHYSICALLY cannot read Marseille data, even via a frontend bug.