Your WMS sees the pick and the dispatch. Everything in between — packing, wrapping, strapping, loading — is invisible. Warehouse IQ recognizes that work from location, motion and sound, with the phone already in the worker's pocket.
A WMS knows what someone typed or scanned, and nothing else. Packing, wrapping, strapping and loading leave no record, so nobody can say where the hours of a shift actually went. Warehouse IQ changes who runs the app: the AI does, not the worker.
The worker works. The app writes it down.
No single sensor knows what a person is doing. Three signals together do. The raw sensor data is not kept: it goes into the AI model on the device, and only the recognized activity is saved.
Small Bluetooth beacons mark the zones: inbound dock, racks, packing tables, the wrapping station, returns, outbound dock.
BLE beaconsThe accelerometer and gyroscope tell walking, standing, circling, lifting and riding a forklift apart.
Accelerometer · gyroscopeThe microphone picks up the tape dispenser, stretch film and pallet truck. It is analysed on the device in real time, and the audio is never stored.
Analysed on deviceThe dot is the worker moving between zones, and each zone fills in with the time spent there. On the right is what the AI recognizes at that moment and which signals it used. Below, the shift builds up as a sequence of activities.
Some warehouses don't want anything sent to a server automatically. For them, Warehouse IQ keeps the whole shift on the phone, and the worker decides what gets sent.
Nothing goes to the backend automatically. During the shift, recognized activities are stored only on the device.
Each activity gets a ✓, or a ✕ for "not accurate" and a pick of what they were really doing. In this example the AI saw wrapping, but it was strapping.
Confirmed activities go into the shift report. Corrections go back as feedback to retrain the model.
Try it: tap ✓ or ✕ on each activity.
Each row is one person, and each colour is an activity the AI recognized. The gaps are picking, moving and breaks, which the WMS already sees through scans. Stack the rows and you get the map of the whole shift: where people actually worked, and for how long.
Every warehouse packs, wraps and loads a little differently. So we don't ship a fixed rulebook. We train the model on your own floor, while your people work as usual.
Together with your team: which activities to recognize, and where the beacons go.
Workers carry on as usual with a phone or watch. Our person on site records when each activity happened.
The model is trained and runs in live operation. We compare what it recognized against what actually happened.
When it gets something wrong and the mistake is corrected, the model retrains on it.
"This motion + this sound in this zone = pallet wrapping."
Motion IQ is already in live use in Slovak hotels as Hotel IQ. A housekeeper carries a phone, there's a beacon in every room, and the AI recognizes the cleaning phases, so each room verifies itself without a checklist. The warehouse uses the same engine with different zones, motions and sounds.
What's needed for the warehouse: train the model on warehouse activities. We've done exactly that once already, in the hotel.
For staffing agencies and their clients: an objective record of what was worked, where and for how long. No more timesheets filled in from memory.
Agencies · 3PLSee where the shift really goes: packing tables, docks, wrapping. Plan staffing and layout on measured hours instead of estimates.
PlanningWorkers see their own day too, and in on-device mode they decide what gets sent. They can check the record and correct it.
TransparencyWarehouse IQ answers one question: what work was done, where, and for how long. It is not a surveillance tool, and it's built so that it can't become one.
Novelty today. Standard tomorrow.