CC32. Cable-Theft Detection & Monitoring
CC32 · Cable-Theft Detection & Monitoring

Know the instant your cable is cut.

CC32 is a compact, cellular-connected sentry for the cable runs thieves target. It watches the line around the clock — and the moment it's cut, it alarms your phone with the location, the evidence, and a police-ready case file.

Instant cut detection Cellular + GPS Push alarms to your phone Fleet dashboard
Explore the system See CCTV evidence →
LIVE UNIT · CC000042● ONLINE
Cable 1Intact
Cable 2Intact
Cellular signalLTE
GPS location52.5741°N, 1.3402°W
Battery92%
Last report2 s ago
The system

One small box. Total cable-run protection.

Everything you need to detect, locate and respond to cable theft — self-contained, connected, and managed from anywhere.

Instant cut detection

Monitors the physical cable line; the moment it's severed, the unit knows and raises the alarm — no polling delay.

Cellular-connected

Reports over the mobile network — no site wiring or broadband needed. It even recovers its own connection automatically.

GPS location

Every unit knows where it is, so you can locate protected assets and pinpoint exactly where an incident happened.

Alarms to your phone

Native iOS & Android apps push the alert the instant it happens — wherever you are, day or night.

Fleet dashboard

A secure web platform to monitor every unit, view history and status, and manage access with role-based permissions.

Battery-backed & resilient

Keeps watching through a power loss, with onboard power monitoring so you know each unit's condition at a glance.

Over-the-air updates

Firmware improves remotely and securely — new features and fixes roll out to the fleet with no site visits.

Relay output

Trigger a siren, light or gate on an event — or power-cycle any external equipment to release it from a locked state — automatically on lock-state detection, on a schedule, or manually.

BLE cable tags

A Bluetooth tag can optionally be fixed, permanently and discreetly, to each cable — two per cable if you prefer. The unit monitors them continuously as a second, independent means of theft detection: if a cable moves away from the station, the alarm fires. And a tag is only registered on the tracking network once an alert is raised — so a stolen cable can still be tracked, with no standing subscription until you actually need it.

Flagship upgrade
CCTV Evidence Capture
On a confirmed cut, capture the scene, read the plate, pick the clearest faces — and hand police a ready-made case file.
See how it works →
Flagship feature · CCTV Evidence Capture

We hand police a case file — not raw footage.

The instant a genuine cable-cut is confirmed, the CC32 captures the scene, reads the plate, and picks the clearest faces — auto-compiled into an incident pack a busy officer can act on.

ANPR number-plate read Best-face selection One-click incident pack Alarm-gated · private land
See how it works View sample case file →
INCIDENT EVIDENCE PACKREF CC32-2026-0714
● TRIGGER — genuine cable-cut asserted · 03:12:47 BST
Vehicle — automatic number-plate read
Number plate VE70 XKP
VE70 XKP0.94 confidence · 34-frame consensus
Persons — best-face frame selection (2 detected)
Person A best frame
Person A
78 px · near-frontal · 6/61 kept
Person B best frame
Person B
54 px · 3/4 left · 4/39 kept

The problem

  • Cable and metal theft is rising, organised and fast — the CC32 already detects the cut in real time.
  • But an alert isn't a conviction. Police are severely undermanned and can't scrub raw CCTV.
  • Handing over 90 seconds of footage almost always ends in “filed, no further action.”

The solution

  • The instant a genuine cut is confirmed, the CC32 captures the scene and does the triage for them.
  • It reads the vehicle plate and selects the clearest face shots automatically.
  • It compiles a one-click incident pack a busy officer can actually action.
How it works

Detect · Capture · Analyse · Deliver

Everything downstream is gated behind a device-asserted genuine cable-cut. Nothing runs speculatively.

1
Genuine cut
The device debounces and latches a real cable-cut alarm — the sole trigger. No speculative recording.
2
Capture
A pre/post full-HD clip is buffered on-device (approach + getaway) and uploaded over the resilient MQTT link.
3
Analyse
The server reads the plate by multi-frame voting and picks the sharpest, most front-on face per person.
4
Deliver
A timestamped incident pack — plate, face gallery, scene still, continuity — ready for police and insurers.
What's in the box

Evidence, not surveillance

Six capabilities that turn a cut alarm into a workable case — while staying strictly the right side of the line.

Number-plate read (ANPR)

Consensus of dozens of frames with a confidence score — recovers a plate that's blurred in any single frame.

Best-face selection

Faces scored on sharpness, size and pose, grouped per person, so police get the clearest shots not 200 duplicates.

One-click case file

A branded PDF: read plate + crops, per-person face gallery, alarm-moment still and full chain-of-custody.

Evidential integrity

SHA-256 verified upload, burnt-in UTC timestamps, admin-only access with every view logged.

Private-land, alarm-gated

Capture and analysis run only on a confirmed crime against your own kit — privacy and proportionality by design.

No recognition, no database

The keeper look-up and any identification are left to police. You supply evidence, not surveillance.

A word on AI

Not all AI is created equal

Most security cameras have built-in AI. That's fine for spotting that a face or a vehicle is present — but it's a small model running on an embedded chip inside the camera, with severely limited processing power.

Once an alarm event is triggered, the associated video is uploaded automatically and securely to our web platform. There it's available to view — and, crucially, it's processed by our AI, Atlas, which in turn calls on a range of cutting-edge, frontier-level AI models. Each has its own strengths and weaknesses, so Atlas uses them all to extract meaningful number-plate, vehicle and facial detail — detecting that a face is present and capturing the clearest possible image of it — to compile into a police incident file. If there is any possibility of recovering a usable plate, vehicle or facial image, these models will find it.

Camera-based AI
A baby
Has just learned to recognise its parent and say 'dada'.
Atlas · our platform AI
An educated young professional
Coordinates the specialists and writes up the findings.
Specialised frontier AI
A world-leading professor
Decades of expertise, one purpose in life: to complete the task you've given it.

Working frame by frame, these models apply correction and enhancement techniques of the kind used in forensic image analysis. Processing a single 60-second clip to this standard takes substantial cloud compute and several minutes — but the results are breathtaking.

ORIGINAL Original low-light frame from the clip
Original frame — the evidence
Untouched, hash-verified. Exactly as captured. This is what stands up in court.
AI-ENHANCED AI-enhanced, corrected read of the same frame
AI-enhanced — the investigative lead
Corrected and sharpened to point police at the plate. A lead, not the evidence.
ORIGINAL Original low-light face frame from the clip
Original frame — the evidence
A face is present, but unusable as-is: dark, distant and motion-blurred.
AI-ENHANCED AI-enhanced, corrected face
AI-enhanced — the investigative lead
Corrected and sharpened into a recognisable image for the police file.
Run in parallelThe same clip goes through multiple frontier models — each strong where another is weak.
Confidence-weightedEvery read comes back with a confidence score, so nothing uncertain is passed off as fact.
Atlas reviews & writesAtlas reconciles the results and generates the police case file — enhanced images as leads, original frames as evidence.

The upshot

The police are handed a concise, factual document showing the crucial information — making enforcement and prosecution far more likely. Once a cable is stolen, it is stolen; catching the thieves rarely brings meaningful recovery. But there is one vital effect — it significantly raises the deterrent. As arrests and prosecutions rise, attacks fall, because the risk to the criminal has been dramatically elevated.

Networks & subscriptions

Low-bandwidth by design

The CC32 connects over LTE Cat-1 — deliberately not a high-bandwidth link to the internet. That rules out real-time video streaming, but real-time viewing isn't needed here, and it isn't wanted: watching a live feed is not what catches a thief.

Instead, on a manual request or when an event fires, only the pertinent clip is uploaded. Given the high resolution of the footage, an upload can take as long as five to six times the length of the clip — and that's perfectly fine, because this is deliberate after-the-fact post-processing, not a live view.

LTE Cat-1, by choiceA lean, low-bandwidth cellular link — no site broadband, and no exposed high-speed connection to secure.
Event-driven uploadNothing streams. Only the clip that matters is sent — on an alarm, or on demand.
After the fact, on purposeThe clip is processed once it has landed, so a slower upload costs nothing that matters.

The upshot — much lower running costs

Because the device never needs a fat data pipe, it runs on a low-cost mobile data plan — keeping the ongoing subscription cost of every unit down, at scale.

The deliverable

Why it wins the sale

  • “We don't just record the theft — we hand the police a workable case file.” That line closes the deal.
  • Turns every unit into an evidence source, materially raising the odds of recovery and prosecution.
  • A concrete, defensible answer to the insurer's and landowner's first question: “and then what happens?”
  • Runs on hardware already in the field — an intelligence upgrade, not a new box.

Open the sample incident pack →

Alarm-moment wide scene, timestamped

Alarm-moment wide scene — real deployments capture full-HD 1080p with burnt-in site, camera and UTC timestamp.

Camera choice

Cutting-edge cameras, plug-and-play

Professional and commercial CCTV cameras are the industry default — and this system fully supports them. But our preferred choice is high-end consumer cameras, for three reasons.

1

A fraction of the cost

Consumer cameras cost a fraction of an equivalent commercial unit — so the budget stretches much further.

2

On the leading edge

The vast consumer market drives the fastest innovation; professional cameras typically run a generation or two behind the latest consumer sensors and processing. Choosing consumer keeps the system current and robust.

3

Plug-and-play install

Every camera arrives preconfigured. Apply power, and the rest is automatic — no on-site setup.

Out-of-hours intrusion detection

Mark a site as time-restricted, and each camera can be set — locally or remotely from the web platform — to raise a warning and upload the clip the moment its onboard AI detects a person or vehicle when nobody should be there. Simple presence detection is exactly what a camera's built-in AI does well; the heavy lifting still happens after the fact on our platform.

More cameras, better coverage

Because each camera costs so much less, you can fit more of them. The more angles and footage you capture, the better the odds of retrieving the identifying detail that makes a police file stick. And it's secure by design: each camera sits on the unit's own isolated network — never exposed to the internet — while the CC32 handles the encrypted upload.

Built to stay the right side of the line. Capture is confined to private land and triggered only by a confirmed cable-cut (or, on a time-restricted site, an out-of-hours detection of a person or vehicle). It is targeted evidence of an offence in progress, not continuous monitoring of the public. No facial recognition and no database matching are performed; identifying suspects and looking up the registered keeper remain matters for the police under their own legal powers. Footage is access-controlled, logged and retention-limited.