Sensorsthe readings that prove the batch was in control.
Cold-chain fridges, cleanroom humidity, filling-line PSI, motor vibration — the readings that prove the batch was in spec the whole time it was running. V5 ingests sensor streams from Bluetooth, LoRaWAN, Modbus, MQTT and OPC UA, applies per-sensor min/max bands, opens a deviation automatically on excursion, and seals the reading window into the batch record so the eBMR / eDHR can be reconstructed minute-by-minute.
Sensors don't just record. They testify.
Temperature, humidity, pressure, vibration — every reading binds to the lot, room and asset that owned it at that timestamp. When an auditor asks "prove the cold chain held for batch B-44871," you don't pull a chart; you pull the chart already attached to the batch record.
Temperature curves. Pressure cascades. Vibration bursts. Door events. All bound at ingestion.
No "exporting CSVs from the loggers" at QA-release time. Whatever shape the signal takes, V5 binds it to the lot, room, asset and operator that owned it at that timestamp.
Bound to the lot in the room. Crossings auto-trigger excursions.
Every signal is bound to a line, SKU, shift and operator before it ever hits a chart.
The dial only tells the truth if the inputs do. V5 ingests PLC tags, counters, reject gates, process probes and condition monitors — and contextualises each one against the order running at that moment. No CSV exports. No spreadsheet reconciliations at end-of-shift.
Splits paid-time into Run, Planned Stop, Unplanned Stop — the A in OEE.
Compared against the standard ideal-cycle for the SKU on that line — the P in OEE.
First-pass yield, tied to lot + operator + changeover — the Q in OEE.
Slow-cycle root-cause: drift below setpoint stretches dwell, eats P.
Auto-releases the line when CIP completes — kills false planned-stop time.
Predictive work-order before failure — converts unplanned stop into planned stop.
Floor signals don't just feed OEE. They explain it — minute by minute.
Every change to the dial points back to the event that caused it: a 47-second micro-stop on INFEED-A, a slow cycle while OVEN-1 zone-2 drifted 6°C below setpoint, a reject burst on CHECKWEIGHER-3 right after changeover. No more guessing why the shift "felt slow."
Every percentage point of A, P, Q decomposes into named events with timestamps, operators and SKUs. Improvement targets become specific work, not vibes.
Ideal-cycle time is held per SKU + line + tooling combo — not a single line-wide guess. Slow-cycle loss is honest.
Stops auto-classify from PLC fault codes; the operator only confirms or overrides on the HMI. Reason-code hygiene stops being a chore.
A 90-second walkthrough of what happens when a probe crosses the line.
Probe reads 8.3°C > spec ceiling. Excursion record created with the exact crossing timestamp.
On-call QA + warehouse lead pinged. Affected lots auto-listed in the alert.
V5 walks back the dwell time; lists every lot in the room during the window with exposure duration.
Deviation auto-drafted with raw probe data + lot list + room + assets — ready for QA review.
A probe's reading is only as good as the chain behind it.
Each reading carries the probe's current calibration cert ID. Out-of-cal data is flagged at ingestion, not at audit.
Gateway buffers locally on network loss. On reconnect, gap is replayed with original timestamps + gap-marker.
MQTT, BLE, LoRaWAN, 4-20 mA via bridge. Bring your own loggers — Sensaphone, Dickson, Ellab, Berlinger, Vaisala.
1-min normal · 1-sec inside an excursion. Storage doesn't drown; investigations get the resolution they need.
Spec bands are versioned with effective-date. Historical data is judged against the spec that applied then.
One click: a signed PDF of the room/asset/lot temperature trace + spec band + cal trail. Audit-ready.
Continuous witness. Bound at ingestion. Calibration in the chain.
Where AI actually earns its place.
AI reads the sensor register the way an engineer reviewing an excursion would — correlating readings across the batch window instead of leaving the reviewer to scroll a 12-month trend line.
- 01
Excursion deviation drafted from the reading window
When a fridge, cleanroom or filling-line sensor breaches its band and auto-opens a deviation, AI drafts the narrative from the actual reading trace — how far out of band, for how long, which batches were running on that line at the time.
Gives QA a populated deviation instead of a blank form and a raw CSV.
- 02
Cross-sensor correlation on investigation
For a yield or quality issue under investigation, AI checks whether any correlated sensor — humidity, PSI, vibration — showed an unusual pattern during the same window and surfaces it as a possible contributor.
- 03
Trend annotation summaries
Across the 12-month trend, AI picks out the annotations worth a second look — repeat excursions on the same sensor, a slow creep toward the band edge — and writes a short summary for the quarterly equipment review.
AI never adjusts a sensor's min/max band and never clears a quarantine triggered by an excursion — those actions stay with the engineer or QA reviewer who owns the deviation.
Just ask V5 — it knows the product cold.
Pick a question or type your own. V5 answers grounded in how sensors & iot readings — temperature, humidity, pressure, vibration, signed to the record | v5 ultimate actually behaves on the floor.
The rest of the platform this plugs into.
V5 isn't a bolt-on. Every module shares the same data, the same audit trail, the same operator. Pick where to look next.
MES
Operator-led execution: scan-gated dispense, step-by-step kiosk, equipment + scale integration, live yield. Built for regulated process & discrete manufacturing.
WMS
Receiving, putaway, transfers, cycle counts, picks, pack & ship — all lot-aware, bin-accurate, FEFO/FIFO-enforced and barcode-driven.
QMS
Built-in QMS: deviations, CAPA, supplier scorecards, in-process AQL, release-by-exception. Aligned with 21 CFR 211/820, ISO 13485 and 111.
Got questions, or want to see it on your shop floor?
Ask V5 — our code-aware assistant — or spin up a workspace. Both are free.