How IncidentSource Turns Radio Traffic into Event Intelligence

IncidentSource is an AI-powered event intelligence platform that converts public safety radio communications into a structured, searchable feed of major events. It listens to radio traffic, uses AI to detect and classify incidents, extracts location and context, and presents everything in a feed with audio playback. If you want situational awareness of fire, EMS, and rescue activity across many jurisdictions without monitoring scanners yourself, this is the mechanism it uses.

The pipeline: from audio to structured event

IncidentSource describes its core function as turning "raw public safety communications into structured major-event awareness." That breaks into four stages.

1. Ingesting radio traffic

The platform monitors public safety radio communications. Its site references Broadcastify audio, and the product framing is "Broadcastify audio in your pocket" — meaning the underlying audio streams come from public safety radio feeds rather than from IncidentSource-operated receivers.

2. AI detection and classification

An AI layer listens to that audio and decides what constitutes a major event. Rather than surfacing every transmission, it filters for incidents worth flagging and assigns a type. In the sample feed, the event types include:

• MVA Extrication
• Shooting
• Structure Fire – Working Fire
• Structure Fire – Multiple Alarm
• Structure Fire – Smoke Showing
• Structure Fire – Defensive Operations
• MVA Major Roadway Shutdown
• Technical Rescue

3. Extracting context and location

For each detected event, the system pulls out the agency and a location string. The feed pairs a jurisdiction with an address or block:

Locations range from precise blocks ("1700 block S 281st Pl, Federal Way, WA") to intersections ("Northwest 53rd Street and Clackmore Road") to "Unknown" when the audio doesn't yield one.

4. Presenting a structured feed

Detected events land in a feed called Recent Major Events, showing the last 20 major events with a 15-minute delay. Each row carries a date/time in UTC, the event type, the agency context, the location, and a View ▶ control that plays the associated audio.

What the platform surfaces at a glance

The homepage also exposes rolling counters, which tell you the scale of what's being processed:

• Signals: 170,447 in the last 24 hours
• Events: 27,997 in the last 24 hours
• Major Events: 768 in the last 24 hours

The gap between "Events" and "Major Events" is the key design point: the AI processes a large volume of activity but escalates only a small fraction to the major-event feed. That escalation step is what the product calls surfacing "what matters quickly."

How to use it

1. Open Radiant Glow, the platform interface referenced on the site.
2. Browse Recent Major Events for the last 20 flagged incidents.
3. Scan the event type, agency, and location columns to find what's relevant to you.
4. Select View ▶ on a row to play the underlying radio audio for that event.
5. Use Sign In for account access, or Get Notified by Squawster to receive alerts. The site also lists Where we cover, What we cover, and APIs as entry points for coverage details and programmatic access.

What to keep in mind

• There is a 15-minute delay on the major-event feed, so this is awareness, not real-time dispatch.
• Coverage is jurisdiction-dependent. The sample feed spans NC, MD, AZ, CA, MO, WA, TX, OH, AL, OK, IL, MA, and LA, but you should check the "Where we cover" section for your area.
• Locations can be incomplete. Some entries show "Unknown," reflecting limits of what the audio contained.
• "Major event" is an AI judgment, not an official classification — treat the feed as a filtered signal, not a verified record.

The practical takeaway: IncidentSource is a filter, not a scanner. It does the listening and triage so you get a short, structured list of significant incidents with the audio one click away.