She has been in the industry for twenty years.

She started as a GIS analyst, learned remote sensing, moved into product, then into strategy. She has watched the sensing layer get built from the ground up — from 30-metre Landsat imagery to sub-metre daily revisit, from manual digitising to AI-powered change detection at planetary scale. She knows what this industry has achieved better than almost anyone.

Last month she sat in a client meeting. A large insurance company. They had just deployed an AI system on top of their geospatial data stack. Best-in-class imagery. Real-time catastrophe feeds. Machine learning models classifying damage at scale within hours of an event.

Someone asked a question she hadn’t heard before.

“When the system flags a property as total loss and the adjuster disagrees — who decides?”

The room went quiet.

Not because nobody had an opinion. Because nobody had designed the answer.

That silence is what this article is about.

What the geospatial industry built — and what it didn’t

The geospatial industry has spent thirty years building the most powerful sensing layer in human history. Satellites that image every point on Earth’s surface daily. Sensors that monitor atmosphere, ocean, land use, and human movement continuously. Models that can detect a building, classify a crop, track a vessel, estimate flood depth — from orbit, at scale, in near real time.

This is Layer 1. And it is amazing.

The industry is now watching Layer 2 arrive. World models and causal reasoning systems that don’t just perceive the physical world but simulate it. Systems that can answer not just where the fire is but what happens next if you deploy resources here rather than there. This is not correlation. This is causation. That is, the difference between a system that describes a situation and one that reasons about it.

The conversations are starting. Benjamin Tuttle, a geographer and technologist working across intelligence, defense, and humanitarian missions, put it precisely in a recent post: world models may be better at reasoning about real-world dynamics, not just generating fluent text. That matters in domains where effective decision-making depends on context, relationships, and change over time.

He is right. And the geospatial community is well placed to shape how that layer gets built, because the sensing foundation it runs on is the one this industry spent three decades constructing.

But there is a third layer. This is the layer nobody is talking about yet.

The most important layer

Layer 3 is the human decision layer. The person who receives the output of the causal reasoning system and has to act on it. The adjuster. The analyst. The incident commander. The underwriter. The intelligence officer.

This is not a technology problem. It is a design problem. And it is the hardest one to put in place.

Here is what actually happens when organisations deploy AI at critical decision points. They build the sensing layer. They add the AI. They put a human next to it. They call it human oversight. And then they discover, usually too late, that they never asked the question that matters most.

What is that human actually authorised to do?

In the insurance meeting, the adjuster had twenty years of field experience. He had looked at thousands of properties. He knew things the model couldn’t encode; the conversation with the homeowner, the detail the imagery missed, the judgment call that only comes from having been wrong before and learned from it.

The system had a confidence score of 94 percent.

Nobody had designed what happens when his judgment and that AI score disagree.

That is not a sensing layer failure. It is not a causal AI failure. It is a human layer failure. And it is being replicated across every industry deploying AI on this geospatial foundations right now – insurance, defense, agriculture, logistics, emergency management.

The question the geospatial industry hasn’t asked itself

The geospatial industry has always been in the business of decision support. Analysts in the middle, translating data into insight, delivering it to the people who need to act.

That model is about to change fundamentally.

Layer 2 systems don’t produce better maps. They produce simulated decision paths. They surface intervention options ranked by predicted outcomes. They run the scenarios before the human has to. The analyst’s role shifts from processing information to governing a system that reasons faster than any human team.

That elevation doesn’t happen automatically. It requires deliberate design. Someone has to ask: what does this person need to know to exercise genuine judgment? What does it mean to challenge the system’s recommendation, and what protection exists for doing so? How do you make the model’s uncertainty visible rather than buried inside a confident output?

These are not questions the geospatial industry has been asked to answer before. They are not engineering questions. They are architecture questions; about the interface between machine reasoning and human authority.

And they need to be answered before the system goes live. Not after the first consequential failure. Not when the lawyers arrive.

Control doesn’t live at the moment of decision. It lives at the moment of design.

What comes next

The geospatial community built the foundation of something it didn’t fully anticipate. The sensing layer it constructed is about to become the perceptual foundation for causal reasoning systems operating at planetary scale. That is a remarkable thing to have built.

But the organisations deploying those systems need something the geospatial industry hasn’t yet produced. Not better data. Not faster models. Someone who understands all three layers — sensing, reasoning, and human decision architecture — and can design the interface between them before the failure makes it obvious that nobody did.

That person needs to be in the room before the system goes live. And in my view that person should be a geospatial expert.

The geospatial industry answered the where. Layer 2 is learning to answer the what next. The question nobody is asking yet is: who designs what the human does with that answer?

That is the gap. And it needs to be solved now, not when the regulatory landscape settles, not when the first lawsuit arrives, but in every organisation deploying AI on geospatial foundations today.

The silence in that insurance meeting was the sound of a design question nobody had thought to ask.

It is the most important question that needs an answer right now.

Matt Sheehan

Matt is a geographer and AI strategist with 25 years at the intersection of geospatial intelligence and decision-making. He maps the architecture connecting three layers most organisations haven’t yet seen together: the sensing layer the geospatial industry has built, the causal reasoning layer now arriving, and the human decision layer nobody is designing.

The third layer is where the value is. And where Matt has his primary focus.

If that layer hasn’t been designed in your organisation yet, that’s usually where the conversation starts: mattsheehan@spatialnext.io