Industrial tech
Real-time signal, from every corner of the plant.
Checkweighers, freezers, doors, dock occupancy, air quality, and production lines, all captured, streamed, stored, and rendered as the live operational picture your floor and your boardroom need. Built for plant-floor reality, not boardroom slideware.
The work
The real-time nervous system for your plant.
Most plant-floor data lives in vendor consoles you cannot reach, paper logs nobody trusts, or one-off scripts that broke the last time IT touched the network. The result: operations runs on what people see and remember, not what is actually happening.
We build the layer above all of that. A single pipeline that captures from machines and sensors (LOMA checkweighers, Ishida packers, freezer probes, door sensors, AQS, dock-door state, production run-rate), streams the events through an industrial-grade bus, lands them in time-series storage, and renders the live picture across Power BI, Microsoft Fabric, Grafana, and operator dashboards. Same architecture for one checkweigher or a hundred sensors across a greenfield site.
We are honest about hardware boundaries. Every dashboard we ship comes with a written "what this would actually require" covering equipment, integration, data, and deployment, because the moment software pretends it has data it cannot get, plant directors stop trusting it. We don't invent telemetry we cannot capture.
Capabilities
What we build inside this practice.
Real-time facility monitoring
The umbrella platform: capture, ingest, transform, store, render. One pipeline handling checkweighers, temperature probes, door sensors, AQS, dock occupancy, run-rate, and anything else on the floor. Live dashboards showing what is happening right now.
Checkweigher integration (LOMA + Ishida)
Live per-pack weight events, full metal-detector contaminant data (Fe / NFe / SST), reject piston state, SPC charts, shift-end compliance reports. Read-only, production-tested driver lineage. See the dedicated LOMA page for the deep technical pattern.
Cold-chain monitoring
Per-zone freezer / chiller temperature probes, door-state events ("door open longer than threshold"), excursion alerts, HACCP audit packet generation. The catastrophic downside, meaning spoiled product plus recall risk plus compliance breach, is exactly why this earns its place.
Air quality & environmental sensors
AQS readings, floor displays, zone-level air quality trends, particulate counts, humidity. Useful for both compliance and operator visibility, because the floor knows when conditions drift before product quality reflects it.
Production run-rate & dock occupancy
Real-time line status (running, idle, faulted), changeover detection, run-rate analytics tied to shift and product. Dock-door occupancy and truck counting for inbound / outbound visibility.
Anomaly detection & predictive intelligence
Beyond simple thresholds: statistical drift detection on historical baselines, "freezer 2 trending toward breach in 3 hours" predictions, correlation analytics across sensors (run-rate vs temperature vs giveaway). Where ML earns its keep, not theatrical AI.
Multi-channel alert escalation
Email, SMS, Teams, and ServiceNow, plus the grown-up version: escalation chains, acknowledge-or-escalate, per-role routing, quiet hours. The system pages the right person, not everyone, and follows up if nobody responds.
HACCP compliance dashboards
Per-zone temperature logs, door event records, contaminant rejects, audit packet exports. Pre-audit ready, structured for retailer compliance requests, exportable on demand. Built for the actual AU food safety standards, not US-flavoured templates.
Manufacturing analytics
OEE, giveaway, reject rate, and downtime breakdown, broken out by line, shift, and product. Tied back to financial impact, not only engineering KPIs. Lives on Power BI for the exec layer, Grafana / live dashboards for the floor.
Roster management tools
Shift-pattern templates, rostering against forecast demand, exception handling. Integrates with UKG Dimensions for time-and-attendance. Closes the loop between roster cost and production output.
Stack
What we work with.
Pragmatic and proven. Edge components run on Windows + NSSM because that's what plant IT can actually maintain. Cloud-side runs on Azure or wherever the customer's existing investment lives.
- Edge: Go event collectors · NSSM-packaged Windows services · Python edge gateway
- Protocols: TCP · OPC-UA · MQTT · Modbus · vendor REST/SOAP · ASCII (LOMA CW3)
- Industrial event bus: Redpanda (Kafka API) for high-frequency sensor streams
- Time-series storage: TimescaleDB (Postgres + Timescale) for sensor telemetry
- Analytic storage: Azure SQL · SQL Server · Microsoft Fabric Warehouse
- Frontend: React · SvelteKit · Three.js (for facility digital twins) · Grafana for ops dashboards
- Backend: FastAPI (Python) · Go services · Node.js where appropriate
- Alerts: email · SMS · Teams · ServiceNow · escalation chains · Power Automate
- Hosting: Azure VM · on-prem Windows · hybrid (edge + cloud sync)
- Integration: LOMA · Ishida · UKG Dimensions · custom OEM · cold-chain / AQS / dock vendors
Pre-sales discovery
One week. We capture what is actually running. You see the gap.
Before any commitment to a full build, we offer a short, fixed-scope discovery: drop in a capture device on the network for a week, then walk you through a written report on what your machines and sensors are actually emitting, and what your current systems are missing.
For a checkweigher line, that's per-pack weight events, MD contaminant counts, reject piston timing, machine identity and firmware versions. For a cold-chain zone, that's per-probe temperature drift, door-state events, comms reliability. For a multi-vendor floor, that's the complete picture across every machine that is actually online.
The report is yours regardless of whether we go further. It's the cheapest, lowest-risk way to find out whether the data you'd need for real-time operations is even reachable on your site, before scoping a build that might be the wrong shape.
FAQ
Frequently asked
What kind of machines and sensors have you actually integrated with?
Plant equipment: LOMA checkweighers (live per-pack weight events, full metal-detector data, reject piston state, SPC), Ishida packers and multihead weighers, custom OEM equipment exposing OPC-UA or Modbus, bespoke vendor APIs. Facility sensors: cold-storage temperature probes, door-state sensors, air-quality sensors (AQS), production run-rate sensors, dock-occupancy sensors. If a machine or sensor can speak TCP, OPC-UA, Modbus, MQTT, REST, or a flat-file drop, we can usually listen.
Can you deliver real-time facility monitoring, not one machine at a time?
Yes. That's the platform we've built. Capture (from machines and sensors), ingest (industrial event bus on Redpanda / Kafka API), transform (per-source protocol handling), store (TimescaleDB for high-frequency telemetry, SQL Server / Azure SQL for the analytic layer), render (Power BI, Microsoft Fabric, live dashboards, operator UIs). The same pipeline handles a checkweigher, a freezer temperature probe, a dock-door sensor, or an air-quality monitor. The data type is just metadata.
We're a food manufacturer. Do you understand HACCP and audit requirements?
Yes. Most of our industrial work is in AU food production. We have built compliance dashboards for HACCP pre-audits, cold-chain temperature and door event monitoring, full metal-detector contaminant logging from checkweighers, and shift-end compliance packets that pass straight to QA. Three consecutive audits passed since one client switched from paper logs to our real-time monitoring.
Can the software live on the plant floor without an internet connection?
Yes. We deploy in three modes: pure local (Windows service via NSSM, edge gateway on-site), hybrid (local-first with periodic cloud sync), and pure cloud where the site has reliable connectivity. Most plant-floor deployments end up hybrid. The event bus and time-series store can run on a single Azure VM or distributed across edge + cloud, depending on bandwidth and latency requirements.
How does the software handle equipment downtime?
Gracefully. We assume machines will go down, comms will drop, and operators will pull plugs to clean. Local buffering, automatic reconnection with gentle backoff (not the bare-RST hammer-loop pattern that has been linked to embedded-stack failures on some machines), and human-readable status indicators so operators know the system is degraded rather than silently lying. Every operational state has a planned UX, not a stack trace.
What about anomaly detection and predictive alerts?
Built in. Threshold-based alerts are the floor: a temperature crossing X, a door open longer than Y. Beyond that we ship anomaly detection (statistical drift on baselines learned from historical data) and predictive flagging ("freezer 2 is trending toward a breach in the next 3 hours"). Anomaly detection uses Power BI's built-in capabilities for the analytic layer, and Azure ML / custom statistical models where the workload justifies it.
What about roster management and shift planning tools?
We build operations-side rostering tools (shift-pattern templates, rostering against forecast demand, exception handling) and integrate them with UKG Dimensions for time-and-attendance. The dashboards close the loop between roster cost and production output.
Related
Other things we do
Data & Analytics
Power BI dashboards, executive reporting suites, and BI diagnostic tools that put decision-ready insight in front of the people who need it.
Explore Enterprise SystemsERP & Systems Integration
Integration across the systems that run your business, plus custom-built connectors for anything that doesn't ship with one. We bridge the platforms your team already relies on.
Explore Product EngineeringSaaS & Product Engineering
We architect, build, and ship SaaS platforms, custom web apps, and mobile tools. Clean code, fixed-scope sprints, and a delivery pace that doesn't drag.
Explore
Software falling over on your plant floor?
Tell us what's going wrong. We'll tell you whether it's fixable, what we'd do, and how long it would take.