Overview
The Analytics Engine is the operator-facing layer of a Frame Analytics deployment. It combines Controller state, telemetry history, alerts, schedules, rules, reports, command lifecycle, and customer-specific modules in one operating context.
The interface should expose what the system knows, how current that knowledge is, what action is permitted, and what evidence exists after the action. It should not merely make stale data look expensive.
Core functions
Observe
Fleet connectivity, data age, current state, trends, exceptions, location, and resource usage.
Decide
Deterministic rules, domain calculations, comparisons, anomaly detection, and assisted analysis where appropriate.
Schedule
One-time or recurring actions with explicit target, constraints, expiry, permissions, and audit state.
Act
Submit bounded operations and track them through accepted, delivered, executed, rejected, expired, or failed states.
Operator workflow
- Confirm scope
Verify customer environment, Controller identity, data freshness, and current equipment state.
- Review evidence
Use current telemetry and relevant history before issuing an operational command.
- Submit an explicit action
Select a supported operation and bounded arguments. Avoid ambiguous free-form command strings.
- Watch lifecycle
Queued, delivered, accepted, executing, succeeded, rejected, expired, or failed are different states.
- Verify the physical result
A broker acknowledgement proves transport, not that a pump, valve, motor, or indicator physically changed state.
Read access to telemetry should not silently grant actuation. Operator roles, command permissions, reauthentication, and audit records remain explicit.
Customer modules
A customer module can add purpose-built dashboards, metrics, device types, reports, schedules, integrations, and analysis while retaining common platform services. This is the normal home for domain-specific logic, equipment vocabulary, and operating constraints.
Availability and stale data
Interfaces expose data age and Controller connectivity directly. A plausible-looking value from yesterday is not current state. Alerts and automation account for stale, missing, duplicated, and out-of-order telemetry.
Rules and assisted analysis
Deterministic software remains the default for explicit calculations and safety-related decisions. Generative or statistical analysis can summarize, classify, or suggest where uncertainty is acceptable, but it does not silently bypass equipment policy, permissions, or command validation.