Keeping false positives in check, with the Institution of Fire Engineers   

Matt Dodwell, PR & Communications Officer at the Institution of Fire Engineers, spoke with Carsten Brinkschulte, CEO of Dryad, about how AI is shaping the next generation of fire safety solutions 

What inspired you to enter fire mitigation and fight wildfires? 

In 2018, the devastating fires in Australia and the Amazon, combined with the Finance Our Future Movement in Europe, were a major wake-up call. My daughter was among the young activists protesting for climate action and that personal connection inspired me and our six co-founders to take action.  

We leveraged our expertise in IoT and telecom technology to combat the wildfire crisis. Wildfires currently contribute to around 20% of global CO2 emissions—up to 8 billion tons annually, equivalent to the entire global transport sector’s emissions. This urgency motivated us to develop a technological solution to mitigate wildfire threats. 

What are the challenges of implementing AI in wildfire prediction? 

The biggest challenge is data. Fire risk is defined by fuel moisture, relative humidity and temperature.

While temperature and humidity can be measured with sensors, fuel moisture is more complex. 

Carsten Brinkschulte

Currently, measuring fuel moisture requires manually collecting samples, weighing them, drying them and weighing them again. There is no automated method available, but we are developing a contactless solution based on lab-tested infrared light reflection to measure fuel moisture. Our early trials are promising and we are now working to commercialize this technology. 

How does Dryad mitigate false positives? 

False positives are a challenge for all fire detection systems. However, our machine learning models, trained on data from over 20,000 sensors installed in Greece, Spain, Portugal, Thailand and beyond, have significantly reduced false alerts.

Initially, false positives were common, but after four years of refining our AI, we have nearly eliminated them.

Carsten Brinkschulte

Each of our sensors now uses AI to differentiate between fire emissions and clean air, ensuring high detection accuracy. 

What makes Dryad’s product unique? 

Silvanet, which has been in development for five years, enables ultra-early detection. Unlike satellite or camera-based systems, which struggle to detect fires beneath dense tree canopies, Silvanet detects fires within 30 minutes of ignition, before the smouldering phase and before an open flame appears. 

Satellites coordinate large-scale responses but are not suited for early detection. Cameras, though effective in some cases, cannot ‘see’ what’s happening beneath tree canopies. Our system, however, uses gas sensors sensitive to hydrogen, carbon monoxide, and volatile organic compounds (VoCs)—the same molecules that give fires their distinctive smell—allowing detection at an ultra-early stage. 

How does the product identify high-risk areas and what data sources do you use? 

Initially, our focus was fire detection, but we are now developing ways to assess fire risk in real time. Our sensors, which cost around €90 each, not only detect fire-related gases but also function as micro weather stations, measuring temperature and humidity—two key indicators of fire risk. 

Do you see AI-driven fire risk assessment becoming a regulatory standard? 

I can’t say for certain, but I certainly hope so. The growing wildfire threat is a public safety issue, and prevention is crucial—we can no longer rely solely on reactive firefighting.

Fire risk reduction strategies, such as prescribed burns to lower fuel loads, must be backed by reliable risk assessments.

Carsten Brinkschulte

AI could play a significant role in ensuring that fire-prone regions, especially those at the intersection of nature and human development (like the Los Angeles wildland-urban interface), take proactive measures. 

Could ultra-early detection devices be integrated into insurance assessments for fire-prone areas like LA? 

We are currently in discussions with various companies regarding this. Fire risk assessment, which depends on factors like fuel moisture, could be invaluable for parametric insurance. AI could enable insurers to assess risk more accurately and encourage mitigation measures that reduce premiums for property owners. 

What is the future of AI in wildfire suppression? 

We are already working on AI-driven wildfire suppression with our new project, Silvaguard. This system integrates AI with autonomous drone technology to detect and extinguish fires before they escalate. This kind of automated response represents the next step in wildfire mitigation. 

As AI technology continues to advance, fire professionals and policymakers must collaborate to integrate AI into comprehensive fire safety strategies.

Carsten Brinkschulte

Whether through improved sensor accuracy, enhanced predictive analytics, or smarter suppression systems, AI is set to play a crucial role in shaping the future of wildfire prevention and management. 

Dryad’s work is just one example of how AI can transform fire mitigation. For organisations developing AI-driven fire safety solutions, whether in predictive modelling, smart detection systems, or AI-powered firefighting tools, the conversation is just beginning.  

This article was originally published in the April 2025 Edition of International Fire & Safety Journal. To read your FREE copy, click here.