System context

Different estates. Same spine. Operators talk to an agent over the estate. Live map, views, and alerts. Not a separate stack per device class.

Control loop

Continuous path from the field to a decision — and back when the estate needs an action.

Platform spine

Every estate lands on this path. New device classes extend ingest and rules — they do not fork the platform.

Connect & ingest

Devices, sensors, vehicles, and drones publish into a common ingest contract. Industrial standards and vendor-specific protocols both land as adapters on that path.

1. StandardsMQTT, OPC-UA, Modbus, HTTPS, and REST for estates that already speak them.
2. Device adaptersVendor protocol documents become ingest adapters that emit a shared telemetry and event shape.
3. Device registryEvery device and asset from first join to retirement — identity, entitlement, and scope.

Edge & cloud

Processing stays on site when latency or offline matters. The cloud holds the durable map, rules, and operator surfaces.

1. EdgeLocal processing when the link drops or the decision cannot wait on the wide area.
2. CloudEvent pipeline, store, metrics & rules, alert delivery, and apps — the durable spine.
3. ScaleFrom a pilot line or site to the full estate without changing the control loop.

Metrics, store, alerts, act

Raw measures become structured telemetry, then metrics and rules operators can trust — with alerts that reach the room that owns the outcome, and command when action is required.

1. Metrics & rulesThresholds, state over time, spatial rules, faults, and estate-specific measures — derived in a reliable time order.
2. Data storeHistory, last status, and read models for live views and reports — scoped by client and role.
3. AlertsMulti-channel delivery with contacts and templates by estate — before the line stops or the asset goes dark.
4. ActCommand and configuration from operator apps back to the device when the estate is entitled.

Agent & surfaces

The agent is how you operate. Live map, views, and alerts. Admin and APIs.

1. AgentChat or talk over plant, fleet, site, and air — answers, alerts, and actions from live data.
2. Live mapStatus, location, and history across plants, sites, mobile assets, drones, and equipment.
3. Views & alertsPlant, building, site, and mobility views. Rules fire when something needs a decision.
4. Admin & configurationHierarchy, devices, zones, signal mappings, alert policy, and reporting.
5. APIsIntegrate upstream and downstream systems on the same platform.

Identity & observability

Shared identity for every surface. Device keys and encrypted traffic in the field. Platform health visible to the team that runs it.

1. Central authSigned sessions, roles, entitlements, and asset or site scope across operator and admin apps.
2. DefendDevice identity, encrypted traffic, and audit for what changed and who asked.
3. ObservabilityIngest health, pipeline lag, alert delivery, and platform status — so failures surface early.
4. UpdateSoftware over the air where the estate supports it — staged and reversible.

Attaching a new device class

A new vendor or device class is an adapter — and new metrics, alerts, and views when needed. Not a parallel platform.

1. ProtocolWhat the device publishes — industrial standard or vendor-specific.
2. AdapterDecode onto the common telemetry and event contract the pipeline already understands.
3. Extend, don't forkNew measures and alerts when the data requires them — agent, map, and admin on the same system.