Heat of the moment

Jay Sadiq explains how his company, FortyGuard, uses AI to understand and manage urban heat.

Q. The world is getting hotter. Why are cities a particular problem?

A: There's a difference between global warming heat and urban heat. Global warming is measured at atmospheric levels. For example, 1.5 degrees Celsius is the global average temperature goal set by the Paris Agreement.

Urban heat is different. It's measured, not in the atmosphere, but near surface levels, about two meters above ground. FortyGuard tries to measure it everywhere, all the time. We're trying to do it every 10 square meters in urban areas, with data sent about every one second.

Temperatures can vary by 10 to 15 degrees Celsius just a few kilometers apart, within the same hour and the same city. Our AI models predict these fluctuations with 89% accuracy.

Q. Why is that important?

If you understand why a certain location is hot, or heating up quickly, then you have some idea of what you can do to cool it down. For example, you might observe that average temperatures in a particular area are still quite moderate, but have risen by eight degrees Celsius in the past few years. To halt that rise and reverse it, you need to understand why it's happening in the first place.

What gets measured gets improved, and the best solutions come from understanding the relevant data. Right now, that understanding doesn’t exist on a large scale.

Q. You don't have sensors sensor two meters above ground, every 10 meters, all around the world. How can you know the temperature in so many different places?

We aggregate temperature data from existing IoT sensors strategically positioned near surface levels in urban environments. This data often spans 10 to 20 years of historical records. After undergoing rigorous quality checks, we integrate it with data from weather stations, satellites, and meteorological models.

To enhance our algorithms, we incorporate additional variables such as humidity, atmospheric pressure, elevation, solar radiation, and other factors. Using advanced data processing techniques, we generate something akin to a "Temperature MRI" that provides an unparalleled view of urban heat dynamics.

This innovation forms the backbone of our AI-driven Large Temperature Models, enabling us to predict temperatures with 89% accuracy at a resolution of 10 square meters in the US. This precision is 115 times greater than existing methods for two-meter temperature data measurement, delivering an unprecedented level of insight at scale.

Q. How has this problem been addressed until now?

A. If a real estate developer wants to understand heat levels at a particular site, it will send a team of engineers there, maybe fly some drones around, collect some satellite weather data. They’ll analyze the data and maybe conclude that a combination of building materials, solar impact, and a lack of shade and vegetation are causing the heat build-up.

This kind of assessment can take anywhere from one to three years.

By contrast, our AI combines measurements, analysis, and insights backed by a lot of data. This lets us formulate a solution that can mitigate temperatures on the ground by anywhere from five degrees to 10 degrees Celsius on average.

Q. What does the customer actually receive from FortyGuard?

A. Usually the customer gets a dashboard with a heat that map that divides areas into grids. The dashboard shows real-time heat data for each tile. You can ask the AI to provide data over different time frames – the past week, the coming year, or whatever you want.

Then you can ask the dashboard why one location is hotter than another. The answer might be, "The hot location is 98% concrete, and only two percent vegetation." Or maybe a certain location is not hot because there’s a lot of cloud cover there. There could be plenty of scenarios and lots of data, but dashboard can engage the user's data to provide customized analysis.

You can correlate heat with air quality, humidity, or other factors. Based on all of that information, you can take the appropriate actions. With Temperature GPS®, logistic businesses can use cooler routes and predict conditions for time- and temperature-sensitive goods. Temperature Realtor® enables property listing websites to offer listings based on micro-climate conditions and allows users to predict energy usage based on location and weather. Electric vehicle makers can use our temperature predictive analysis to increase battery range, efficiency, and durability. And a home energy application provides heat-predictive analytics to optimize climate control and energy efficiency.

Q. Are you building a consumer app for this?

A.  Early next year (2025) we'll introduce a "freemium" dashboard, similar to Chat GPT. The free version will have some basic capabilities allowing you to map the temperature (at your desired locality). If you want something more advanced – analytics, for example, because you’re an engineer – that will be available in the paid version. Later, we’ll create an app.

Early next year, we are also launching beta-version APIs that let developers integrate temperature data into their software. Our mission is to make temperature data a standard feature integrated with everyday applications. So, maybe you don’t need our dashboard, but you want to put temperature data on your dashboard. If you have software that looks at distribution of energy across a city, you can add a temperature overlay to the that data.

Q. What kind of customers do you work with?

A. To date, we've worked mostly with developers, cities, and national ministries of energy or urban planning. We’ve worked with Masdar in Abu Dhabi. They want to build a cool, walkable city in a hot country. We collaborated with their engineering team to suggest ways to lower the temperature.

We've learned that different customers have different needs. Their main goal might be to lower the overall temperature of a particular area. But they might also have additional goals, such as reducing energy consumption to cool down building, improving retail sales or lifting property values. Using artificial intelligence, we can advise on all of these things in seconds, straight from our dashboard.

Our APIs will be sold through channel distributors or by partnering with existing platforms such as NVIDIA Market Place, Google Cloud Marketplace, and Amazon Web Services, as well as through embedded integrations with platforms such as Google Maps (temperature routing), Zillow (micro-climate listing), Autodesk (design optimization), and NVIDIA Omniverse (layer of heat for digital twins).

Q. What makes you optimistic about the future of cities?

A. Urban heat can be reversed. Cities use different building materials, they're located in different climates – near the ocean or far from it, at low elevations or high ones. There's a range of factors at play, and that can give you something to work with when you’re trying to make them cooler.

Q. Conversely, what keeps you up at night?

A. The "boiling frog" syndrome: the idea that a frog placed in a pot of boiling water will immediately hop out, while a frog placed in cool water that heats up slowly will eventually boil to death.

This may not actually be true, but it illustrates a point: heat is rising but we keep adapting instead of solving the root cause of problem. We think of things like building bigger malls with better air conditioning, so that people can come in out of the heat. Cities aren’t solving the problem because they don’t have the data. If they did, they would take action and jump out of the boiling pot.

In the coming five to 10 years, the world's biggest challenge will be rising temperatures. This will affect our economy, our health, and our prosperity. For many people, from the time they wake up in each morning, their biggest challenge will be trying to stay cool.