
Live alerts.
Shows each confirmed incident with how serious it is, where it is on the map and a short text summary.
- Map pin & severity
- AI-written summary
- Fresh-first sorting

We watch live camera video, spot crashes, and send alerts with location, video clip and how serious it looks—all in seconds. Made in Pakistan for roads everywhere.
For traffic police and motorway control rooms. Works with Safe City and motorway cameras you already use. You do not need to buy new cameras first.
Indeed, in need—the minutes right after a crash are what we care about most.
How it works

Connects to live video from motorways, city CCTV and other feeds you already run.

Looks for real collisions on the video and tries not to flood staff with false alarms.

Another AI step reviews the moment, judges how serious it looks and writes a simple incident summary staff can read fast.

Confirmed incidents go out to your operators and emergency teams with map location and supporting video.
Built for
Fits the kind of CCTV and motorway camera setups already used in Islamabad, Punjab and Sindh. Adds crash detection on top—you keep your existing cameras.

Islamabad capital region

Punjab province

Sindh province
Capabilities
Traffic staff and emergency teams use one screen instead of jumping between many tools.

Shows each confirmed incident with how serious it is, where it is on the map and a short text summary.
Track each case from open to closed. Staff can see who did what and when.
See where crashes happen often so you can plan patrols and fixes.

Add many cameras and check that each feed is healthy so broken streams do not go unnoticed.
Send confirmed alerts by WhatsApp and email with a small preview image so teams see it quickly.
Different login roles for admins, operators and analysts so people only see what they should.

Many cameras, one picture
From the field
Below you can watch a short demo and browse images from pilot setups: crash detection on roads, operator screens and hotspot maps. Some clips may be blurred or staged where privacy requires it.
In a live deployment your IT team can also connect camera feeds using RTSP or RTMP, alongside normal files or hosted video—whatever your security rules allow.
Screenshots







The Platform
Three layers in one product: watch video, judge incidents and notify teams—because crashes need answers in seconds, not IT tickets.

Computer vision tuned for real roads here—bright sun, dust, rain and mixed vehicles.
A second AI step reads the clip and writes a short severity note so staff do not guess from raw video alone.
Sends confirmed alerts to your operators and Rescue 1122 with location and supporting footage.
Numbers above are design targets. Real pilots vary with internet speed, video quality and how many cameras you connect.
The People
Graduates from Air University, Islamabad. Started as a university project and grew into a company focused on faster roadside response.

Ph.D in computer vision. Faculty supervisor for SafeRaah at Air University, Islamabad.

BS Data Science, Fall 2022, Air University. Leads the team, product direction, and delivery.

BS Data Science, Fall 2022, Air University. Builds and ships the SafeRaah software stack.

BS Data Science, Fall 2022, Air University. Data pipelines, labelling, and analytics behind the models.

From classroom to company
Supported by the Air University Business Incubation Center (Cohort 4). Local mentorship and funding help us turn research into software cities can actually deploy.
Get in touch
Cities, motorways, campuses or fleets—tell us what you manage and we will plan a small pilot on cameras you already have.
You keep ownership of your feeds and rules. We add crash detection on top.