Telemetry infrastructure for physical systems.
Capture, normalize, and transport machine data from edge to cloud—reliably and securely. Visibility first; then diagnosis, prediction, and decision support grounded in real behavior.
What Ironlink is (and what it isn't)
Ironlink is a telemetry infrastructure layer: capture signals close to the asset, normalize, buffer at the edge, forward securely for dashboards and alerting. Not a single-dashboard project or a fragile integration web—a durable foundation that scales.
Edge telemetry that survives reality
Deterministic behavior through outages. Buffering and sync for unreliable links.
Vendor-agnostic by design
Mixed fleets, partial signals. Optional PLC/SCADA; not required for value.
Security first-class
TLS, authenticated brokers, least privilege, production hardening.
How it works
Edge gateway ingests sensor/device signals, publishes normalized telemetry over MQTT. Buffered locally, forwarded to cloud when connectivity allows. Dashboards and alerting now; instance baselining and decision support later.
1) Sense
Temperature, power, duty cycle, vibration, acoustics, ambient—whatever is economically observable.
2) Edge gateway
Normalize, timestamp, buffer, forward. Unattended operation, predictable failure modes.
3) Cloud ingest
Secure retention, dashboards, alerting. Clean event stream for inference and recommendations.
Why Ironlink compounds
Features are easy to copy. Performance improves when a system accumulates grounded reality and learns from outcomes. Ironlink is designed around that loop.
Longitudinal behavior data
Time-aligned histories under stress, maintenance, failure—drift over months. What makes "health" measurable.
Instance-aware inference
Per-asset baselines: this machine in this environment, not generic assumptions.
Outcome-labeled feedback
Alert → action → outcome. Cause–decision–effect turns monitoring into decision support.
Signal without invasive access
Most machines don't expose deep internal states. Ironlink infers condition from observable signatures and non-invasive measurements—the shadow a machine casts.
Tier 1: Observational
Reinterpret exposed signals: variance, recovery time, drift, stability—control signals into diagnostic signals.
Tier 2: Adjacent sensing
Clamp-on current, external temp, vibration, acoustics, ambient. Non-invasive sweet spot.
Tier 3: Embedded/OEM (optional)
PLC/SCADA/historian or vendor feeds where available. Valuable, never required.
Roadmap (incremental hardening)
Each milestone tightens guarantees and expands scope while staying compatible.
Visibility baseline
Signal capture, edge buffering, operator dashboards on a real asset.
- Telemetry + normalization
- Local MQTT + metrics
- Machine-state dashboards
Edge production
Repeatable deployment, unattended operation.
- Runbooks + health checks
- Auth, TLS, secrets
- Failure-mode handling
Secure transport & cloud ingest
Edge→cloud with TLS + auth, bridging for intermittent connectivity, cloud dashboards.
- Cloud MQTT broker
- Bridging/sync
- Cloud alerting
Fleet + decision support
Fleet onboarding, instance baselining, health scoring, outcome-grounded recommendations.
- Device lifecycle
- Health scoring
- Human-in-the-loop
Request a pilot
Site with physical assets and want grounded visibility? Share the basics and we'll propose a scoped pilot.
hello@ironlink.co.za
Include: site, asset types, connectivity, signals available, outcomes that matter (uptime, quality, safety, energy, maintenance).
We support PLC/SCADA where it exists; default is vendor-agnostic sensing and behavior-first inference.