Spatial data – a record of physical or virtual data – is important to a variety of industries, yet a gap remains between collecting the raw data and gaining AI insights from it.
I recently had the opportunity to speak with Damian Wylie, the head of products at spatial ETL, analytics and GeoAI company Wherobots, about the challenges of working with spatial data. This conversation has been edited for length and clarity.
Q: What was the problem you saw with gleaning AI insights from spatial data?
A: Let’s first start with what spatial data is, and then we can drill into some of the problems. So spatial data is a record of places, objects or activities, say, in a virtual or physical space. A virtual space could be something like a Metaverse or a game or an application. We’re going to spend most of our time today talking about the physical space. The physical space is anything tangible. This could represent things above our atmosphere, in space or in deep outer space, or could also be things on the ground or even below ground. Spatial data can represent trips, routes, land, roads, a road network, parcel data, crops, building data, and so on.
Q: What are some of the kinds of industries that rely on this data?
A: This data is fundamental to a variety of various industries, from mobility, agritech, insurance, energy, telecom, retail, logistics. And what companies want to do with this data is they want to build better products, better services and make better decisions. There are small-scale use cases all the way up to large scale-use cases. So if you’re a company that’s maybe making decisions around where you’re going to place your retail store, that’s an example of a type of organization like, maybe a Starbucks. Or, there are companies trying to figure out where to invest in their next solar panel farm, or a commodities company trying to understand what the value of certain crop types are going to be this year.
Q: So what is the gap that exists between collecting this raw spatial data and being able to gain AI-ready insights from it?
A: The primary challenge that developers generally face when trying to work with this data is, they look around the landscape of options. The tooling out there is not purpose-built for the end application, which requires the developers to have to build workarounds. You look around the ecosystem, you’ll see a number of extensions that are added on to support spatial data. And that’s a lot of complexity that the developers have to undergo. Developers are trying to put this very complex or noisy data into those systems and expecting to get some output out of it, with some amount of performance or even at a cost that’s reasonable. So there’s really some economic challenges that developers or companies face today with respect to putting spatial data to work.
Q: How is Wherobots addressing these challenges?
A: We believe that when someone can take your idea about the physical world and bring it to market and bring it into production, within minutes rather than weeks or months, that’s going to unlock a lot of innovation. There are remote sensing applications that we are working on, and that’s a growing area of interest within the market, because a lot of companies want to put these sensors to work that are assigned to drones and satellites. So you can imagine those satellites and drones are flying around areas of interest, where maybe you’re scanning rivers, for example, and then having the systems and tooling that makes that very economical to use. The market needs lower cost, much more performance and easy-to-use tooling.
Q: How does your platform make that data AI-ready for developers to use.?
A: The computing systems we’re talking about are like databases, big data analytics systems. You’ll see that those systems have evolved to support, but they weren’t inherently built for, spatial data, and so the bottlenecks that exist in those systems will surface through at a higher cost to the customer, while delivering sluggish performance. We’re also working on this full stack, because when someone’s working with the spatial data, they’re not just interfacing with the computing system, they’re working with storage systems, and they’re also working through development interfaces.
Q: How will AI agents improve use of spatial data?
A: When you look at LLMs today, what they’re trained on is the internet, but the internet is not providing a first-party representation of the physical world. It’s generally inferences, derived from news articles and other data points online. So if you were to ask an LLM, for example, “How fast is this fire spreading,” or, “What’s the area of that fire,” it would go to the web for an answer. We believe it is possible and will be possible, to make AI agents capable of working directly with physical world data to answer a whole new category of questions that people just aren’t using LLMs for.
So, what we see happening is, yes, there’s an explosion of data there, and there are many use cases for that data, but there’s a big gap in the middle between the use cases and the data itself.
