Closed beta
TetherX AI
Traditional motion detection tells you something moved. TetherX AI tells you what moved. Video AI identifies people, vehicles and animals. Audio AI classifies 63 security-relevant sounds. Both run entirely on the TetherBox - no footage leaves your site.
Intelligence on the edge - the AI runs on-device, alongside normal recording. No bandwidth impact, no video leaving the premises, no separate analytics server.
Runs On-Device
A recent-generation TetherBox handles the full AI workload locally, in real time, with no separate server to install or maintain.
Searchable Footage
Every detection becomes metadata. Filter recordings by what was seen and heard, not just by when motion occurred.
Two Senses
Video AI and audio AI run together on the same TetherBox - audio covers the corridors and spaces a camera cannot see.
Privacy by Design
No video or audio is sent to the cloud for analysis. Only classification metadata is stored. The raw footage stays on-site.
We Taught TetherBoxes to See
Not just detect motion - actually understand what is in the frame. Video AI identifies people, vehicles and animals in real time, tracks their movement across the scene, and classifies what it finds.
Detection That Holds Up in the Real World
The images here are from a live beta site in Canada - night time, snow on the ground, well below zero. Two people detected at distance as tiny figures in a wide scene, at high confidence. Most camera analytics would not register them at all.
Each detection is tracked frame to frame, so you see the full trajectory: where someone came from, the path they took, and where they went.
Search Your Footage Like a Database
Instead of scrubbing through hours of recordings, filter by what matters.
Person, Vehicle or Animal
Narrow results to the detection type you care about.
Clothing Colour
Top and bottom classified separately for precise filtering.
Vehicle Colour
Find that specific car across an entire day of footage.
Moving or Stationary
Filter out parked vehicles, or focus on loitering.
Direction of Travel
Which way were they heading when they crossed the scene?
Trajectory Tracking
The full path of every detection, mapped across frames.
Show me everyone wearing a red jacket between 6pm and midnight. Results in seconds.
We Taught TetherBoxes to Hear
Video analytics have a blind spot - they only work where the camera can see. Audio AI classifies 63 distinct sounds across 7 categories, running continuously as a second sensory layer.
Detecting What Cameras Miss
Glass breaking behind a wall, a smoke alarm in an adjacent room, someone screaming off-screen - none of these trigger video analytics. A site with 16 cameras still has corridors, stairwells and spaces with no visual coverage.
Audio events appear with distinctive purple icons on event thumbnails, instantly distinguishable from video detections when an operator is scanning dozens of events. Loud sounds above 99 dB SPL are automatically flagged regardless of classification.
Filter by category or individual sound class, and route notifications by type - glass break alerts to the monitoring station, dog barking to the site manager.
Seven Categories, 63 Sound Classes
Emergency
Screams, cries for help, shouting
Alarms
Smoke detectors, fire alarms, security sirens
Intrusion
Glass breaking, door forced, wood splintering
Vehicles
Engine start, horn, tyre screech, reversing beeper
Voice
Speech, arguing, crowd noise
Animals
Dog barking, bird of prey
Weather
Thunder, heavy rain, high wind
individual sound classes across every category
Two Sensors Create Context Neither Has Alone
A person detected on camera at 3am is interesting. A person detected at 3am with glass breaking on audio is an incident. Video AI and audio AI run on the same TetherBox, at the same time, with no extra hardware - each filling in what the other cannot reach.
More Analytics Coming in Q4 2026
Video and audio AI are live in beta today. The following detections are targeted for release in Q4 2026, shaped directly by integrator and customer requests.
People Crowding
Flag when too many people gather in an area, for managing occupancy and spotting unusual build-ups.
Unattended Counter
Alert when a till or service point is left without staff for a set period of time.
Visitor Counting
Daily footfall counts from a camera covering an entrance.
Weapons Detection
Detect a range of weapons in the frame. Accuracy depends on camera placement and scene-specific training.
Face Recognition
Match faces against a known list, with a positive-match API for access and alerting workflows. In active development.
Need Something Specific?
Much of the roadmap comes straight from integrator requests. Tell us what your customers need.
Roadmap features and timing are indicative and may change.
The Intelligence Stays On-Site
TetherX AI runs entirely on the TetherBox. No cloud processing, no footage leaving the premises, and no separate analytics server to run.
No Extra Server
Nothing to install and no separate box to maintain. It runs alongside normal recording.
No Bandwidth Impact
The AI workload never touches your connection - it is processed where the cameras are.
Data Sovereignty
For sites with strict data requirements, the video and the intelligence both stay on the premises.
Cloud Backup Still Works
Optional cloud recording continues exactly as before, per camera.
What You Need
TetherX AI requires an AI-capable, recent-generation TetherBox. The box handles the video and audio workload locally, in real time, with no additional hardware and no change to how footage is recorded or backed up. If you are not sure whether a site's TetherBox is AI-capable, your TetherX account manager can confirm.
Full Details & Hardware RequirementsJoin the TetherX AI Beta
We are running a free closed beta for security integrators. Test video and audio AI on real sites, give honest feedback, and help shape the product before general release.
