As wildfires become more frequent and increasingly affect regions with little previous experience of large-scale fires, effective wildfire management is becoming an urgent priority.
From detection and monitoring to emergency response, the challenge is no longer simply collecting more data, but ensuring that critical information reaches the people who need it, when they need it.
In this interview, Guy Schumann, Founder and CEO of RSS-Hydro, discusses the changing demands facing wildfire management and the role of Earth observation, geospatial technology and AI in supporting disaster response.
He explores why existing technologies don’t always reach firefighters in a useful format, and why solutions need to be designed around established emergency protocols rather than expecting responders to adapt to new technology.
Schumann also discusses the importance of collaboration between technology providers and firefighters, the role of human expertise and trust in developing effective tools, and the need to improve access to wildfire information for underserved communities.
Looking ahead, he argues that faster development, funding and deployment will be essential as wildfire risks continue to challenge communities and critical infrastructure.
Wildfires are increasing, and many regions are experiencing them for the first time. How important is it that governments and communities prepare for this changing reality?
It is actually very sad because the wildfire season is not over. I’m calling you from a country that has no history of wildland fires until this season. We had small fires, but now everybody in Europe has fires. I heard today that Germany has two federal states dealing with fires.
Scotland has also been in bad shape. They even called the Disaster Charter for help with the fires because the peatlands were burning. They have the same problem as Canada, where fires can go underground.
Because peatlands are so carbon-rich, the fire can continue burning underground without flames and then come back to the surface when it reaches vegetation. It is very ugly.
I think people need to get prepared for this to repeat and, unfortunately, accept that this may become the new reality we have to live with.
Of course, we could also help nature avoid getting to that state, and I’m not convinced that we can’t change things. But even if we assume for a moment that we can’t, we still need to be much better prepared.
We need to start living with the assumption that this is the new norm. If we do that, I don’t think we’re going to die from this in the future. But right now, it looks like these issues are becoming political questions very quickly in Europe.
So it is more difficult than simply saying, “Let’s get prepared.” For individuals, preparing their private homes is one thing, but how do you regulate an entire system of preparedness without some decisions becoming political issues? It is very complex.
With increasing wildfires, detection, monitoring and response are becoming increasingly important. Where are the biggest gaps in how wildfire information is currently collected, shared and used on the ground?
If you’re thinking about monitoring and detection, I would say the technologies already exist. Collection and perhaps prediction of how a fire will spread are not yet at the level required. Maybe the technology or development isn’t there yet, but monitoring and detection are probably there.
The bigger problem is what happens with that information afterwards.
The Earth observation, geospatial, technology and AI communities often assume that everybody who needs this information should be able to log into a platform and look at it, or perhaps receive it directly on their phone. That’s great, but ask a firefighter with a hose in his/her hand who is trying to put out a fire. None of those methods necessarily work for them.
They may need to know every five minutes where the fire is right now. Imagine a firefighter who has just put out a fire one mile away five minutes ago. They need to know: has it come back or not? Because they’re still close enough to go back and put it out again.
We have the technology to monitor that. But the way we deliver it to the end user on the ground is often wrong.
If we’re targeting managers sitting in an office, a platform might be fine. But even then, they need to know how to access it. Perhaps that manager would prefer a report as a PDF, or even a piece of paper on their desk. At worst, it might come by email.
I’ve sat in a few emergency rooms and said nothing because, when you see how these operations actually need to happen, you realise we’re not delivering information in the right way.
So the challenge is not necessarily the availability of data, but turning it into actionable information that can be used under pressure?
Exactly. We have the information, but how do we get it to the firefighter on the ground?
I think it needs to be delivered through whatever communication equipment that person already carries. If they use a radio, the information needs to be delivered through the radio. If they use a pager, it should arrive as a message, perhaps like a traditional SMS, rather than through an app on an iPhone.
They may have traditional equipment because it works under harsh conditions. It might not be able to consume more than 150 characters of text, but it can beep and give them a message.
We can put information from space into that kind of system. We can do it. So we need to change our thinking and ask: what information is actually helpful, and who are we delivering it to?
We cannot ask a firefighter fighting an active fire to change the way they operate. If they are a manager in an office and their protocol means they receive information by email, then that’s where the information needs to go.
I think it is wrong to push people to change how they operate during disasters. You can’t do that. It is protocol, much like defence. You cannot simply change the protocol. You have to fit the protocol.
In the past, we might have said, “Look at this great technology, but you need to have a Simple Storage Service (S3) bucket on Amazon Web Services (AWS) to get the information.”
Nobody in field operations or search and rescue is going to activate an S3 bucket unless they have enough experts on the team to do that. An active search-and-rescue team with a boat and a dog is not going to activate an iPhone to look at an S3 bucket on AWS.
It sounds logical when I say it, but that’s really what is happening in reality. We are serving data into these buckets because it is automated and technically possible, but it isn’t necessarily having an impact because the people we want to help aren’t operating that way.
So, to answer your question, I don’t think there is a major gap in detection and monitoring. The enormous gap is how we feed that information to the ground.
We shouldn’t push technology forward in the way we think it should work and then expect the end user to adapt. We should adapt the way we deliver it to an end user who already has an operational protocol.
The same applies to a utility company supplying water or energy. You can’t tell them, “I’ve put it in an S3 bucket in Amazon,” if their protocol only allows data to come through a very particular highly secure VPN into a platform built for them by a major company. You have to feed the information in that way.
If they say the best they can do is accept a CSV or Excel file, then you need to put your information into the Excel file; otherwise, they will never consume it.
You’ve spoken about the importance of turning complex data into actionable intelligence. Has your experience in emergency situations changed the direction of RSS-Hydro or the way you approach this process?
Yes, it has. Fortunately, IT technology has also changed at the same time, particularly with Large Language Models (LLMs) and AI.
I’m not an AI person or a machine learning (ML) person, although perhaps I understand machine learning more because I studied a lot of statistics. But AI and all of this is relatively new to me.
These agentic systems can do a great job, but they also need data in a form that machines can interpret. A map or image is much harder for these systems to interpret than simple text-based or numerical information.
It is actually very similar to what a person in search and rescue needs. An LLM can’t necessarily take very complex image data and interpret it for decision-making. It is much easier for a machine to receive parameters in text or numbers and say, for example, “fire intensity: 200.” That is easy for a person and easy for a machine to interpret.
Even easier would be: “Fire intensity: high.” Easier still would be: “Fire intensity: evacuate.”
That is effectively an order from a command centre. You don’t need to worry about interpreting the fire anymore. You simply follow the order and get people out. For an LLM, the principle is similar.
My experience in these emergency rooms told me that we need to completely change the way we look at geospatial data when we are dealing with disasters and emergencies.
We can no longer simply feed them a gigabyte-sized image. We also can’t feed them a gigabyte-sized interpretation of that image. We need to condense the information into a message.
Doing that properly is difficult, which is why we work closely with stakeholders at the end of the process, including firefighters and disaster managers. We ask them what parameters they want and what the resulting unit should be. Is it a text, a number or another unit?
We then develop that capability in a way that we believe can serve a wider market segment rather than just one customer.
We will not patent the methodology behind how we think about actionable information because, if it is genuinely solving a problem and working, we can still make money by being faster and better than everybody else delivering the service.
But we won’t file a patent on the methodology because it would almost be unethical. If you find a cure for cancer, you can patent it for a number of years, but eventually it becomes freely available. In disasters, we need to move in the same direction.
Wildfires are not only a threat to lives and communities. They can also damage critical infrastructure. How important is it that this becomes a greater consideration in disaster preparedness?
It is extremely important.
Whether it is a wildfire, flood, earthquake or another disaster, the impact on critical infrastructure is always a concern on the defence side. But it also needs to become a concern in civil society.
In the US, it is already treated this way in some respects. When a hurricane hits the coast, it can be considered almost equivalent to an attack because the infrastructure impact can be enormous.
A 200-kilometre-per-hour windstorm combined with a five-metre storm surge may not kill the same number of people in a well-prepared society, but it can hit infrastructure in a very similar way.
In Europe, we need to rethink how defence funding can be combined with civil funding in certain circumstances because wildfires will hit critical infrastructure.
There may be a pipeline somewhere that nobody knows about because it is a secret pipeline. Every country has infrastructure like this carrying fuel or supporting other parts of society. There may also be internet cables or other infrastructure that absolutely cannot be damaged.
But, given what we’re seeing from nature, that infrastructure may now be more likely to be damaged by floods, fires and earthquakes than by a foreign entity.
If that is the case, we need to find ways in Europe to combine defence and civil funding for certain areas.
There is currently a push to increase defence budgets, but it can be difficult for companies like ours to access that money if we are not doing traditional defence contracts. We want to do more to protect against natural disasters, including floods and fires, while also protecting critical infrastructure.
In the US, when a hurricane hits, access to significant funding can suddenly become available to protect a city against flooding. Why? Because critical infrastructure is involved, and that can unlock other budgets.
I don’t know how this works in the UK, but in Europe, defence budgets aren’t necessarily released if there is suddenly a wildfire in Luxembourg.
Should they be? I think absolutely. If a wildfire gets close to something that needs to be protected for defence reasons, then the fire is a threat to national security.
Do you think that, if we don’t act soon, damage to critical infrastructure could become inevitable?
Yes, and even more so now given the geopolitical tensions everywhere. You can easily imagine how someone or an entity could use what nature does to their advantage.
It would be very wrong to assume that next year will be much better for wildfires than this year. We can’t simply say, “This year was bad, but next year won’t be.”
That is a very dangerous assumption.
What if it gets worse? I’m not saying it will, but what if it does?
We need to act on the “what if” much better than we have in the past. The probability of those “what if” scenarios is increasing, and we need to start acting on them.
We should stop thinking about this as something that will happen 100 years from now. It is already getting worse. We have experienced it this year.
People in France have been talking about this on national television. They have interviewed elderly people and asked whether they have ever experienced heat like this. They have seen individual days above 40°C, of course, but not prolonged periods like this.
We should assume it will get worse before it gets better.
How essential is collaboration between technology providers, wildfire management organisations, firefighters, emergency services and land managers?
I think it is absolutely critical.
It is not easy because there is a huge gap between the two segments of society: technology and solution providers on one side, and the people fighting disasters and responsible for saving lives on the other.
There is a huge disconnect, but there are very good reasons for that. Why should these groups necessarily be connected? Their missions are very different.
But if those missions are becoming more closely connected because firefighters need better information, faster information and predictive information, then there absolutely needs to be a connection and a dialogue between them.
For example, our FireSENS application is about to be verified by one commune in Luxembourg. Fortunately, the person has firefighting experience.
We are serving fire intensity every 15 minutes for every ongoing fire in an administration, country or continent. What we are not necessarily good at is knowing what that fire intensity actually means in terms of what is burning.
If a fire reaches one intensity level, it may be burning grass. At another level, it could mean the forest is burning. At very high heat intensity, it could potentially be burning fossil fuel or be an accident fire.
Firefighters know this.
You can give them the joules of the burning energy and they can tell you whether it is grass burning. They can even distinguish between fresh grass and dry grass.
That is their expertise.
So, by working with firefighters to classify fire intensity according to what is actually burning, we can improve the information we provide.
You could say, “Just put it into Claude, and it will tell you.” Yes, perhaps. But if I go to a firefighter and say, “I’ve developed this classification,” they will ask, “Who have you been working with?”
If I say I’ve been working with Irish firefighters and we’re selling the service to Luxembourgish firefighters, they might say, “Absolutely fine. We all work the same.”
But if I tell them I developed it with an LLM, I think they would question it, even if the result had the same accuracy.
We should never lose the human connection because it is based on trust.
Even if Claude gives me exactly the same result as a firefighter, that result would probably still be much better received by a Luxembourgish firefighter if I can say that I have spoken to their commander and the classification is based on the Luxembourg system.
That would be the best outcome, but it means talking to people in each country.
Even if I said I had spoken to a Moroccan firefighter and they had helped develop the classification, the Luxembourgish firefighter might still accept it because we operate according to broadly similar protocols in search and rescue around the world.
We should never expect someone to pay for a service simply because we have developed it with an LLM. You can still use the technology, but you should never forget the human connection.
The more technology you use, the more important it becomes to connect with the end user you want to serve, whether that’s a client, a paper mill wanting your technology or a firefighter.
You need that customer connection.
Is this something RSS-Hydro is looking to do more of in the future, collaborating with firefighting and emergency teams to inform how you present your insights?
I think we would love to do that more. We love educating and being educated, and we love making those connections.
It is just, as I said, very difficult to do that and still run your business. It is not easy to make those connections because they are all based on trust and relationship building, and these things take time.
But we would love to do that, and we are doing it. It takes a lot of human connection and, especially now, trust takes a lot of time to build.
How important is it to increase access to wildfire technology and information for communities that may be underserved or lack infrastructure?
I think it is very important.
With the increase in technology, and if you are smart about how you use it, you can build tools and connectivity systems that allow you to serve people through different pricing models.
You can start serving people according to their budget.
If there is an underserved but highly vulnerable community that says, “We’ve only got €500 to spend this year, but we really value your service,” you should be able to provide something basic that genuinely helps them.
It can operate on the same principles as your more advanced service, but at a much more affordable price.
Then you have companies and customers with larger budgets who want an advanced service and can pay for it.
You can even have a free tier, but it would obviously provide only the basic version. I think that basic version needs to be very good. It shouldn’t be a free trial that runs out after three months.
You can’t do that with early warning systems or crucial information about an ongoing fire.
You should be doing the opposite.
You can have major companies paying significant amounts for very good services while offering a free service to public firefighters around the world.
Nowadays, I don’t think anybody can say that this is impossible.
Companies with public valuations of $2 trillion would have been unimaginable 10 years ago. Now it happens.
So money is not necessarily the problem.
I think there is an ethical responsibility to offer a free service if a company is making enough money to do so. And, to be fair, major technology companies already give many things away for free. There are countless apps and services that people don’t directly pay for because somebody else is paying.
That model should be adapted and adopted for disasters.
Why should firefighters in the field, fighting an active fire, have to pay for a subscription?
But should a reinsurer pay for the service? Absolutely.
Payment also brings responsibility. If a major customer pays for a service, they expect it to be reliable, perhaps 99% available, and to help them adjust their own models.
That makes sense.
A major reinsurance company also wouldn’t necessarily want a free contract because that could mean there is no liability on the service provider’s side.
So get yourself a major customer and then use that model to provide an ethical, free service to people fighting disasters on the ground.
I’m talking about disasters, not defence.
What would a truly connected wildfire information ecosystem look like, and what are the key steps to achieving it?
The first step is to identify your stakeholder or anchor customer, whoever you want that to be in the future.
Get in touch with them. Find out what they need and where their pain points are. Get the requirements and understand their needs, then build your system around that.
But ask them everything.
How do they want the information? How quickly? How often? How many refreshes do they need per hour?
You also need to establish minimum acceptable criteria.
They might initially say, “I want the best possible resolution and information every two minutes.” But you need to ask whether they would still be satisfied with something less.
For example, if you provide five-metre resolution, would they still be happy?
This is something we learned the hard way. You need to ask these questions so that people don’t simply say, “I want the best of the best.”
Establish the minimum acceptable criteria and develop towards that.
Then make sure that the customer is actually willing to pay for the service.
If your stakeholder is a firefighter and you want to provide the service for free, make sure you have another anchor customer that can pay for it.
Don’t build something with a single customer that you already know cannot pay for it.
Those customers are still extremely valuable because they give you visibility and make you feel valued. But you also need a larger customer that can ask for an advanced service and is able to pay for it.
After all, we are a for-profit company.
Then maintain that connectivity with the customer. You can develop the service in an agile way and add features in the future.
I don’t think it is clever anymore to say, “That technology doesn’t exist, so we can’t do it.”
We are in a world where almost everything is possible now. But we must never forget the human in the loop, particularly when it comes to end users and customers.
You need to have a customer partner that understands that, once the service is developed, they need to pay for it.
If they can’t pay, find somebody else alongside them.
We have found that to be very effective, and sometimes more effective than relying on research funding.
Your major anchor customer can take the risk and co-develop the service with you. That means the development money comes from the customer rather than entirely from research funding.
Research funding is great money, and you can do a lot of innovation with it, but it can take two years to access that money. There are proposal calls, tenders and applications.
We operate in that world too, but we have two pillars in the company. Research funding allows you to innovate significantly, but it can take too long to build momentum around something that you want to have a real-world impact.
We need to get faster.
Everybody needs to spend money a little faster because the technology world is moving faster, and nature and disasters are catching up with the speed of technology.
The one thing that isn’t keeping up in many places is the speed of cash distribution.
It is still very slow, whether it is money from customers or funding agencies.
We need to speed up that process and make it easier to write contracts.
We are now getting into the habit of telling customers we trust: “Don’t worry. We’ll deploy the service first.”
If we have already signed something, we don’t wait for every final detail. We simply say, “Okay, we’ll serve now.”
Why? Because by the time everything is set up legally under traditional timelines, the wildfire season may already be over.
Things need to catch up.
We need to become much more efficient because nature is moving very quickly right now.
We are in an El Niño year, so there are some countries that may unfortunately be very hard hit by floods between November and March, particularly around monsoon seasons.
If a government comes to us now and says, “We need your service because we want to be ready for November,” then we should deploy it now.
We should get a commitment letter signed, but we shouldn’t wait until every contractual and legal detail is complete before providing the service.
November is too soon for that.
We need to commit, and payments need to happen faster.
We’re also a small company, so cash is always important. But I mean this more broadly because technology is moving so quickly that we are even seeing it with students.
If you go to university and complete a master’s degree in AI, you had better finish it and get a job because, in two years, AI may have moved so quickly that your master’s becomes irrelevant.
It’s scary, but unfortunately that’s the reality.
We need to move at the same speed, or slow everything down. But there are some things we can’t slow down.
Unfortunately, the increase in wildfires is currently one of them.
