Find the leaks.Then keep finding them.
An integrated leak detection service that combines high-sensitivity IoT sensors, CNN-based acoustic analysis and LLM reporting with a field crew that verifies every AI-flagged point. We run the survey, pinpoint the leak and hand over a repair list — then leave permanent sensors on the network so new leaks are caught as they appear.
Why a service, not just a device
Leak detection is a workflow, not a sensor.
Most utilities own listening equipment. What they lack is the daily routine: collecting, scoring, verifying and closing out every suspect point. Our service supplies the equipment, the AI and the people, so the routine runs without adding staff.
Skilled listeners are scarce
Expert acoustic survey depends on a few experienced technicians, and their judgement varies with the operator, the surface and the hour of the night.
Surveys are campaigns
A one-off survey sees the network for a few nights a year. Between campaigns, the first signal of a new leak is usually a resident’s complaint.
Alarms without a verdict
Sensors that only raise alarms push the diagnostic work back to the utility. Without field verification, false positives erode trust in the data.
Data that is never reused
Survey results stay in spreadsheets. The leak history that could map risk across districts is never assembled.
How the service runs
Five steps, repeated every day.
We deploy sensors district by district, score every recording the same day, verify in the field and report — and the permanent loggers keep the district under watch after the campaign ends.
Deploy sensors
Sonic GL loggers on meters, valves and hydrants across the DMA; Sonic M2 units in valve chambers for permanent watch.
Collect
Overnight acoustic and pressure recordings upload through the field app and the LTE-M network, with over 99% data collection.
AI analysis
Recordings are converted to spectrograms, noise-cancelled and classified by CNN and Auto-Encoder models; adjacent sensors are correlated to position the leak.
Verify in the field
Our technicians survey every AI-flagged point with a ground microphone within 72 hours, and the verified result is fed back into the model.
Report and repair
The LLM agent issues the daily report and the repair list with positions; the utility or our crew repairs, and the sensors confirm the leak is gone.
What you get
Deliverables the utility can act on.
| Deliverable | Frequency | Format |
|---|---|---|
| Daily analysis report | Every morning | E-mail · web dashboard |
| Verified leak list with positions | Weekly · on confirmation | Map · spreadsheet · repair order |
| Alarm on new or worsening leak | Immediate | E-mail · SMS |
| District leak-risk map | End of campaign · annually | Map layer · PDF |
| Sensor data access | Continuous | NELOW web and mobile, no software to install |
| Campaign review and next-year plan | End of campaign | Workshop · report |
Reliable in harsh environments
IP68 sensors, over a year of battery life and an LTE-M network optimised for underground chambers keep data flowing from the field.
Fewer false alarms
Pressure and leak sound are analysed together, with pipe material and ambient noise built into the model, so alarms mean something.
Nothing to install
Web and mobile access for the utility’s staff, and natural-language questions answered by the NELOW agent.
Insight for next year
A year of data becomes a district-level risk map and a prioritised plan for the following campaign.
Service options
Three ways to engage.
From a single survey campaign to a contract paid out of the water it saves.
DMA leak survey
A time-boxed campaign over one or more DMAs: sensors, AI scoring, field verification and a repair list.
Permanent watch
Sonic M2 loggers stay on the network and the platform reports daily, with our team verifying what is new or getting worse.
NRW reduction project
A multi-year programme paid against the verified reduction in non-revenue water.
Find out what your network is losing.
Tell us the DMA and we will come back with a survey plan.
