IoT — cold-chain monitoring with 42 freezer sensors

The scenario

You’re the night-ops manager at NorthChain Warehouse, a cold-chain 3PL storing frozen product for 40+ brands. You have 42 freezers across 4 rooms, each one publishing temperature every 5 minutes. If any freezer goes above -15°C for more than 30 minutes, your customers’ product is at risk and your contracts have financial penalties.

Right now you don’t have monitoring — you have someone checking a spreadsheet export every morning. That person is you, and you’re tired.

This tutorial uses a CSV export as the starting point. Once it works, you’ll switch to live MQTT (see Next steps).

Four questions:

1. Is any freezer out of spec right now? — current-state check.
2. Which freezer has been trending warm — not broken yet, but heading that way?
3. Are any sensors dead — no readings in the last hour?
4. Can we alert on all three — 24/7, route to Slack + SMS overnight?

Download the sample data

Drop both into a chat. On Free plan, telemetry.csv (~8 MB) imports in under a minute.

The Agent surfaces: F-0017 has been intermittently spiking above -15°C for 3 days — not consistent enough to show on the “latest reading” view but clearly pathological when you look at the trend. Likely a failing compressor or a defrost-cycle bug. Worth scheduling maintenance before it fails hard.

Keep it: ＋ Save as daily Report — mornings, you get the week’s warming-trend list.

The Agent ranks all 42 freezers by reading-count shortfall. F-0032 shows 0 readings in the last 18 hours, which was already caught in step 1 — but this query would also catch partial outages (readings every 15 minutes instead of every 5, e.g.).

Keep it: ＋ Save as Script — reusable weekly sensor-health audit.

The Agent creates the Watch with three sub-conditions, each routed differently. Test-fires it once — F-0017 and F-0032 trigger in the test.

Keep it: ＋ Build Dashboard — a shop-floor screen showing live temp per freezer, color-coded by room. Open on the control-room TV full screen.

What you built in 25 minutes

• 1 Report — daily warming-trend watch.
• 1 Script — sensor-health audit, weekly.
• 1 Watch — three-way alarm (critical / warning / sensor issue) with severity routing.
• 1 Dashboard — live shop-floor screen.

You now have 24/7 cold-chain monitoring. Your night shift doesn’t check a spreadsheet; they get a Slack ping only when something’s wrong. The sensor-offline case used to take hours to catch — now it’s in your Slack within 60 minutes.

Next steps: switch from CSV to live MQTT

The CSV version of this tutorial is the static-data path. To run it live:

• Point your devices at Tablize’s MQTT broker. See IoT, MQTT & cameras — credentials, topic format, QoS.
• Re-ingest as a live table. Your telemetry table starts filling as devices publish. All the Reports / Watches you built on the CSV keep working — they query the same table name.
• Enable push for the Dashboard. Swap the 5-second refresh for MQTT push — sub-second updates on the control-room screen.
• Add the Spatial UI. Space → Room → Freezer → Temperature Point. Tablize’s Asset Graph gives you a clickable map of every freezer with live state, time-travel slider, and camera overlay if you wire cameras too.

Simulator — if you want to test the MQTT flow without real hardware, scripts/docs-samples/iot-simulator.py (coming soon) replays the telemetry.csv data onto a local MQTT broker at real-time speed.

Next steps in this industry

• Read IoT, MQTT & cameras — full reference, including camera feeds and time-replay.
• Read the Manufacturing tutorial — IoT in a factory setting, same primitives, different use case.
• Explore Federation — if you run multiple warehouses, each workspace owns its own data; federation gives you a single roll-up view.

Nearby tutorials

• Manufacturing — machine telemetry in a factory.
• Logistics — when what you store needs to get shipped.