This project needed more than a normal script. It had to scan Singapore reliably, avoid duplicate alerts, handle API and Telegram failures calmly, and send timely accident updates into a Telegram group without someone staring at a dashboard all day.
The project needed to monitor accident alerts across Singapore without relying on a person to keep refreshing a map. A single broad query was not enough because location data can be uneven, duplicated or noisy. The system needed structured coverage and sensible filtering.
We built a PHP automation that calls Waze-sourced alert data through the OpenWeb Ninja API, scans 10 overlapping Singapore regions every minute, filters for accident alerts, removes duplicates, stores state and sends formatted Telegram group messages with a Waze map link.
The finished system delivered 419 accident alerts in a recent 24-hour operational snapshot, with 0 Telegram delivery failures recorded in that snapshot. More importantly, it turned raw external data into a workflow people could actually use.
A lightweight worker runs every minute so the system can check fresh traffic information without manual action.
The monitor checks 10 overlapping Singapore regions instead of trusting one broad request to catch everything neatly.
It reads Waze-sourced alert data through the OpenWeb Ninja API and normalises the useful fields for processing.
The system keeps Singapore accident alerts and ignores unrelated alert types, non-Singapore items and incomplete data.
Incident IDs are remembered for 7 days so the Telegram group does not receive the same alert again and again.
Formatted alerts are sent to the Telegram group with location details and a Waze link for quick follow-up.
On paper, an alert bot sounds simple. Call API, send message, done. Then real life comes in wearing slippers and says: duplicate alerts, missing coordinates, API timeouts, region boundary issues, Telegram failures and state tracking. So the system had to be boring in the best way: predictable, defensive and easy to monitor.
The build avoids unnecessary dependency weight and keeps the moving parts understandable. It uses concurrent scanning to reduce delay, a local state store for deduplication and retry metadata, and operational checks so we can see whether the worker is still healthy.
The Telegram message focuses on what matters: incident type, location, time and a map link for fast action.
Many automations look impressive in a demo. The real test is what happens at 2am when the API is slow, Telegram rejects a request, or the same accident appears through multiple regions. This system was designed with those small annoying realities in mind, because small annoying realities are usually where business systems break.
Send Telegram, WhatsApp-style or email alerts when important business events happen.
Pull external data, clean it, check rules and notify the right team without manual refresh work.
Track jobs, incidents, service areas, branches, delivery zones or map-based activity.
Replace repeated checking, copying and chasing with a system that handles the routine parts.
If your team is still checking dashboards, copying messages, monitoring inboxes or refreshing portals manually, there may be a calmer way to build it.
Founder-led web, mobile app and custom management systems team. Send us your issue, current website, or idea and we will help identify the practical next step.
Small fixes, audits and consultation can start from a practical entry point. Full websites, apps and systems are quoted after scope, screens, roles and workflow are clear.