Strength in Partnership: Pioneering AI Fire Detection with FireScout AI

Over the past three years, WatchTowers Networks and Alchera X have worked closely to bring FireScout - an AI smoke detection system headquartered in California and developed in South Korea - into the hands of emergency services through WatchTowers Command. What began as a promising concept is now a trusted tool that helps teams detect fires faster and more accurately, and in places they might otherwise miss.

As any firefighter will tell you, fires rarely behave how you expect them to. They can spark in the middle of the night, deep in remote terrain, or on days when no one expects them. Thankfully, FireScout AI stays switched on. It watches camera feeds 24/7, using smart algorithms to scan for smoke in real time. 

FireScout AI’s effectiveness is amplified by its integration with the WatchTowers Command platform, which provides high-quality, consistent camera imagery through our Sentinel Stations. Our open platform architecture offers a stable and adaptable foundation, ensuring that AI systems like FireScout perform reliably and consistently across diverse agencies. 

Raw CCTV feeds alone often fall short of what’s needed for robust early detection - it’s the synergy of superior infrastructure, continuous calibration and real-world deployment through WatchTowers that brings AI potential to life. By integrating AI technologies in an agnostic manner, we enable a transparent and unbiased evaluation process, tackling challenges like alert fatigue and unclear performance insights. This ensures emergency services receive accurate, actionable alerts they can trust, enhancing their ability to respond swiftly and effectively to fire threats.

What sets it apart isn’t just its reach - up to 80 kilometres from a WatchTowers station - but how reliably it performs. In over 1,000 fire events, FireScout AI has delivered consistent, timely alerts that help responders act before things escalate.

Real-world testing and smarter detection

That reliability was earned through rigorous real-world testing. With support from the Minderoo Foundation, we put FireScout and other AI systems through their paces, testing them in real conditions against real fires. We weren’t interested in hype - we wanted proof. FireScout AI stood out, especially at night and over long distances, and kept false alarms to a minimum.

False positives - when a system thinks it sees smoke but doesn't - can slow everything down. Crews are sent out to check, only to find nothing. That wastes time and energy. FireScout AI tackles this with a simple but powerful approach: every time a fire is confirmed or dismissed, the system learns, using the information to improve its accuracy over time. That learning loop made a big difference during our two-year testing phase, where we saw the system get noticeably smarter.

Bringing FireScout AI into WatchTowers Command wasn’t just a plug-and-play job. The two systems had to be carefully aligned so they could share data securely and respond quickly. Now, FireScout AI analyses imagery from WatchTowers cameras in real time and sends alerts when smoke is detected, which automatically kicks off the next steps inside Command. That might mean flagging a hazard to local teams, queuing up camera views, or activating a wider response depending on what’s needed. It all happens quickly, without manual work.

The collaboration worked so well because Alchera X shares our approach. They weren’t just handing over software - they rolled up their sleeves. We solved challenges together, focusing on simplicity for the end user. While we work with multiple AI detection partners, FireScout AI stands out for how smoothly it integrates and how well it performs in the field.

What it means for emergency services

FireScout AI has now been used in more than 800 real-world detections across Australia, with similar success overseas. One example comes from Sonoma County, California, where the system identified smoke ten minutes before anyone called emergency services. That kind of head start can give incident controllers the critical time they need to get ahead of a fast-moving situation.

For emergency services, FireScout AI reduces the need to monitor dozens of feeds manually. Instead of having staff constantly scanning for signs of smoke, the system does the watching and only sounds the alarm when there’s something to check. This frees up teams to make informed decisions, rather than getting stuck in round-the-clock surveillance.

These kinds of results are only possible when technology is built to serve the needs of the people using it. That mindset - one of adaptability and collaboration - is core to how we design everything at WatchTowers. We don’t build closed systems or force agencies to adopt just one way of working. WatchTowers Command brings together multiple data sources, detection tools and camera feeds into one place. That means agencies can use the best mix of technologies for their region, their workflows and their teams.

What's next

FireScout AI’s deployment in Australia through WatchTowers Command helped demonstrate what AI fire detection can offer when it’s part of a robust, real-world system. Looking ahead, we’re continuing to evolve the WatchTowers ecosystem to ensure detection services are fast, reliable and resilient. That includes working with new partners and models, which gives agencies more options to explore new use cases and other applications of the technology.

Our experience with FireScout AI showed what’s possible when detection is backed by stable infrastructure and deep operational integration. That’s what this partnership has always been about: working together to prove what’s possible, then building on those lessons to make the system even stronger. As the demands on fire detection grow, we’re continuing to evolve - with partners who are ready to meet that challenge.