Geospatial AI for Climate and Nature – San Francisco 2025

[Music] Well, I'm really honored to have Diego Seaz Gil here from Pachchama. Pachchama is one of the world's leading companies working on very high quality nature-based carbon credits. So, maybe just walk us through your core business model, but then we'd love to hear about some of your other uh commons based initiatives. Yeah, thank you for having me. A pleasure to be here.

And yeah, Pachama started six and a half years ago with a mission of helping restore nature and help solve climate change. Um and our insight was that uh this explosion of remote sensing data that uh was happening right with more satellites put on orbit, more uh sensors like LAR becoming uh available um and of course with AI uh advancing exponentially that those technologies together could allow a whole new way of measuring, reporting and ultimately bringing awareness and hopefully funding to projects that are restoring or conserving nature uh which is critical to solve climate change. Um so fundamentally we are a geospatial AI company. Um we started building these models uh again six and a half years ago. We spend a lot of time building the infrastructure to be able to ingest uh satellite data from NASA, from the European Space Agency, from private companies and then to fuse that data with LAR data collected by airplanes uh around the world and then field plot data collected by you know teams around the world and then train algorithms to come out with predictions right so for example one of the models that we built is a model to estimate carbon on forest um using uh the height of a trees as a proxy, right?

And um so we use all that data to help companies confidently invest in these nature-based carbon credits projects. And as someone said before, carbon markets are not perfect, but it's what we have today. A couple of billion dollars that flow every year from corporations into projects. Historically uh the market was full of uncertainty um and frankly sometimes even bad practices by people who could took advantage of methodologies that were kind of like very flexible. And our intention has been to in one hand uh help companies see uh more clearly uh the impact of these projects and then help project developers um faster develop their projects by having access to the data that is needed to report and comply with the methodologies.

It's been a journey from a model business model perspective. We've been evolving from a brokerage model to a originator model and now back to a platform with services fees model where we're trying to really reduce any conflict of interest and um be able to serve both the corporates, the investors and the projects in the different stages of their journey. Um and yeah fascinating journey in the sense of like learning so much by doing and you're working to open up some of the core geospatial AI technologies working with industry consorcia tell us about that yeah so we last year we uh shared two scientific papers one on the canopy high model that I mentioned before where we explain how we combine satellite data with LAR data collected by airplanes, but also with lighter data collected by the International Space Station, a program called Jedi, where we have a global map of uh dots uh with the height of the trees and the topography of an ecosystem and and we've been working with that data set for a long time. We came to demonstrate that you can uh with a good degree of confidence estimate uh carbon stocks and then use that to look at the history of an ecosystem and even predict what's going to happen with that ecosystem in the future. And and and we are getting together with other players in the market.

You could call competitors, but actually there isn't competition. We were trying to protect the forest and the planet and and it's been great to see that, you know. So we've been collaborating with companies like Planet and Chloris Geospatia and Space Intelligence and and others on sharing notes, sharing data and and there's still uh an opportunity to make it a more formal data commons that everybody can access. But I think it's fundamental that that we make that data available and and those algorithms available for everyone uh to be able to, you know, act. Well, that's very much in the spirit of the Buckminister Fuller Institute.

We're very interested in data commons, knowledge commons at bio regional to planetary scales because we should be competing to regenerate as quickly as possible in the most inclusive way with the deepest participation by indigenous communities, local communities. Let's sort of compete about that. And a really exciting example is a project called cooling the climate. So this is one of our Buckminister Fuller Institute design lab projects. Cooling the climate is working broadly with many other organizations to look at restoring the Amazon basin.

And what you're doing could be immensely helpful if we can open source the geospatial data, what what's happening on the ground, the canopy, but also the algorithms equally important. And the significance of the Amazon basin is that if we could even restore to historical levels of forest canopy, we could probably reduce global mean temperature by up to, you know, one degree Celsius, which could be up to 2/3 of the current impact. So I'm curious as we think about some of these large initiatives. So symbiosis for instance, you were an integral part of starting an industry effort investing billions of dollars in in reforestation specifically. How can we team up?

How can we team up to do the big things like restoring the Amazon, like uh making sure the Congo basin is protected, the great rainforest of Indonesia, even those three key tropical forests, if we can make progress there, we could actually have a chance to stabilize the planet. Yeah. I mean I've been uh very inspired by all the previous conversations about you know radical collaboration that is needed and yeah you're pointing to an example uh you know last year we were talking to so many big corporates here in the Bay Area and hearing similar questions and and concerns. So, we actually organized a dinner here at Greens, uh, one of my favorite restaurants in San Francisco, and and we invited the the biggest tech companies that that we knew, and we said, "Look, this is a time for collaboration, for bringing together uh technical teams and and and financial teams and to co-create something that can actually bring uh this world of nature-based carbon markets to the next level. And from that actually, the symbiosis coalition came to happen.

Uh today uh four core members Google, Microsoft, Meta and Salesforce have committed to invest over a billion dollars into a new generation of nature-based carbon projects. Uh we are their tech platform receiving hundreds of submissions of projects around the world and using the geospatial data uh to help them rapidly move uh them through uh diligence and to funding. Right? Because another problem here was, you know, how long it took to get certification and access to financing and every planting season that you miss is a missed opportunity to, you know, get trees on the ground and get carbon sequester. uh but it's been fascinating to see um you know this symbiotic uh emergence and then all the ecosystem that is forming around that coalition because then now there are vendors data providers uh consultants are coming to collaborate and it's really cool to see um and I hope to see more of that so what's emerging is almost like a digital twin of the earth so Nvidia has a major earth digital twin project.

What you're talking about with these data coalitions and and sharing algorithms, you know, points to almost what Fuller called the world game, right? This ability to have very diverse stakeholders play in person, virtually, you know, whatever the setting and actually begin to visualize regenerative outcomes. And of course we can start with core protected areas restoration opportunities but there's no reason we can't even bring this into neighborhoods cities and begin to have digital twins of bio regions. That's another one of our design lab efforts. So I'm I'm curious um what you see as the potential for this kind of digital twin work at at different scales as a fractal uh visualization process.

I mean that is ultimately the vision that we're working for. I think that the vision that Baky had back in the day is finally feasible. Uh and you know with all this exploding remote sensing data that is only going to continue to increase uh and you know the compute power that we have today which is also you know growing rapidly and new algorithms, new techniques uh new infrastructure. Um, I think that this vision of a simulation of the biosphere uh and the atmosphere and the geosphere where we can start playing out a lot of different possible scenarios for the future of Earth and based on that decide what to do uh as stewards of uh spaceship earth, right? um that vision is finally I I can see that we're almost there.

Actually, I think that is a super underinvested space in AI. So much money going to large language models, not enough going to large earth models, which is what I am hoping to see um come soon. But there's many companies working on it and I think that it's not going to be something that is going to be built by one single company. Uh it's going to be layers of collaboration. is going to be a centralized and common uh effort.

Uh there is a lot of different companies doing things. You know, of course, Google has been building Google Earth Engine for a long time. Super important contribution. Um you know, Nvidia is working on the digital twin and uh and there are many others. Um but I think that that's something that I hope that more people get excited to help build.

And it's important to recognize it's not just static visualization. It's it's gamification. It's scenario building. It's actual modeling and simulation. So imagine if we could better connect our existing climate models with the amazing geospatial canopy vegetation models that you're building.

And this is really a cutting edge for climate modeling now where it's very hard in current models to understand the benefits that forests provide for uh nucleation for clouds. So, you know, trees emit certain chemicals that help clouds form, and the the chemistry, the physics is really complicated and interesting. So, we're kind of at the very edge of what our models can do. And if we could combine that with new regenerative economics where in insurance companies, for instance, would realize if we could invest systemically in large-scale restoration, let's model it out. we can say not with precision for one event but we can say parametrically we can reduce the chance of the next terrible LA fire for instance.

So another BFI effort is looking at restoration work throughout Southern California to reduce the fire risk in LA because that was the ultimate cause of the incredible speed and dryness of those Santa Ana winds. So combine that with rehydrating the LA basin itself. So, we're at this sort of cutting edge where we need to think about massive like trillion dollar investments globally in these restoration efforts. So, I really hope that we can team up and curious what what other key industry partners or nonprofit entities you would want to involve. I would love to collaborate on this, but yeah, I think that weather modeling is going to improve dramatically.

It's already happening. I do want to plug in two startups that spun out of Pachama. Um, one is actually by my former co-founder Thomas who's building a uh climate insurance company using better machine learning models to estimate um, you know, short-term weather um, and and then with that help farmers get better insurance and then from farming he's going to move to real estate and other things. But but I think that there's a great opportunity for improving insurance. Um another startup that came out of Pajama is a company called EarthScale and they're building the infrastructure to do the ingestion uh visualization and processing of m geospatial machine learning models which again we we took so much time to just build the infrastructure to then do the the modeling that we wanted to do.

So they're doing that as a service for others. Um but yeah, I mean I think that as you said when all these intersects with an economic system that then can create incentives um for farmers to grow the right crops in the right place for you know uh land owners to do reforestation and conservation with the right incentives compared to the alternative. Uh when cities can know where to build and where not to build. Um, I think that that is, you know, hopefully the the application that we want to see for these models. So, last question for you.

There's an incredible company called Quarksala that has started working in British Columbia where I grew up. So, the heart of the coastal temperate rainforest. Brilliant. Their idea was to take what's called a tree farm license, which is a provincial, you know, permission to log pretty much as much as you want, and to actually acquire that and turn it into a right to regenerate license. So, they were able last year to change the legislation so entities could buy the license, but then use it only for regeneration.

and they've set up a majority indigenous owned company, Guacalo, with tokens and incredible monitoring that probably could benefit from all the open source technology that you're talking about. So, my hope is that we're going to have a massive shift from extractive economic models to regenerative. That needs to happen alongside all the incredible tech. So, as you look ahead, you're running a company. Are you beginning to think about the next few years and how maybe we'll start to have incredible momentum to do things like restore forest versus it being so uphill like like Troy told us about earlier.

Yeah, I mean I think that uh for companies uh there are moments of flaring and moments of focus. That's something I actually heard from Robin from Planet. I don't know if he's still here. Um but you know we've been going through moments in which we expand and we say we want to do all these things and then we're like okay no let's focus on the things that we actually can be the best in the world at uh right now we are in a moment to focus but yeah when I look at the future uh there's so many things that that that can be done we know biodiversity is such an important thing it was said before carbon is just one single uh you know metric of ecosystem value but but there's so many more things that that we should measure and account and value for um current credits are just a very you know uh imperfect mechanism that today drives billions of dollars but I think in the future we need to keep innovating mechanisms to give value to you know regeneration um and and yeah I'm super excited about what I hear from others uh you know both on the crypto community on the AI community and I think that uh even though we are in this moment of political uncertainty. I also try to remember to think in cycles, right?

And cycles come and go, the pendulum go one one direction, one and the other. And and I think that, you know, the next few years might be uncertain, but in the midterm, in the long term, the opportunity to build a a sustainable civilization is there and we have to work for it now. Beautiful. Thanks so much. Thanks for your work.

Yeah. You [Music]