What Is Spatial AI in Construction?

Definition: What Is Spatial AI?

Spatial AI is a type of artificial intelligence built to understand real physical spaces. Most AI tools today work with text and documents. Spatial AI works with images, video, and location data from the real world.

In construction, spatial AI means systems that can read what is happening on a job site and get the right information to the right person at the right time.

Jeevan Kalanithi, CEO of OpenSpace, is one of the clearest voices on this topic. OpenSpace has spent eight years building the visual capture technology that spatial AI needs to work on real job sites. His Foundamental University masterclass explains what spatial AI means for construction, why it is different from general AI, and where it is headed.

Watch the full masterclass here: university.foundamental.com/masterclass/jeevan-kalanithi

Why General AI Is Not Enough for Construction

Most AI tools are built around language. They read text, write text, and answer questions based on what exists on the internet.

Construction does not run on text. It runs on physical reality — materials, structures, and spaces that change every day. A language model can read a spec sheet. It cannot look at a job site and tell you whether the work in room 603 is on track.

As Kalanithi puts it: "All the AI developments that everybody's been reading about are fundamentally oriented around language, text, and documents. They aren't necessarily great at understanding real physical reality."

Spatial AI fills this gap. It is trained on visual data from real job sites, not text from the internet.

How Spatial AI Works in Construction

OpenSpace started with a simple idea: job site teams needed better visual records, and capturing them had to be easy.

Their product lets site teams record a full visual snapshot of a job site just by walking through it with a 360-degree camera. No extra steps, no new workflow. The capture happens during a normal site walk. The processing happens automatically.

Over eight years and thousands of projects, OpenSpace has built up a large dataset of job site images. That dataset is the foundation for their spatial AI work.

The AI capabilities built on top work at two levels.

At the insights level, project managers and executives can track schedule progress, spot delays, and get productivity reports — all generated from visual data, without being on site.

At the field level, things get more interesting. "Imagine someone in room 603 trying to solve an issue," Kalanithi says. "Spatial AI agents can flag issues relevant to that person based on their location, and even suggest who should resolve it." OpenSpace calls these field agents. They know where you are, what that space looked like last week, and what needs attention right now.

Why Location Matters for Spatial AI

Knowing what happened on a job site is only useful if you know where it happened.

"If I don't know where something happened, I might as well not know it at all," a customer told Kalanithi.

This is what makes construction different from most AI use cases. A chatbot does not need to know where you are standing. A field AI agent does.

OpenSpace built a technology called Auto Location that tracks the position of a smartphone indoors without GPS. This means every annotation, every punch list item, every flagged issue gets tied to a specific location. That location data is what makes the AI useful in the field rather than just in the office.

What Spatial AI in Construction Is Not

There is a lot of AI hype in construction right now. Most of it is about rendering tools and visual outputs.

Kalanithi is direct: "A lot of the generative AI potential currently used in AEC has been stuck in the 2D world of visualization. I've suffered through too many conference presentations where the only thing you get is a new workflow for producing renders."

Spatial AI is not about renders. It is about understanding what is actually happening on a job site.

There is also a hard reliability constraint. "Buildings need to stand up and not kill people. Using AI systems that are right 95% of the time is not feasible in construction, because the 5% of the time they're wrong will skyrocket your insurance premiums."

This is why OpenSpace builds human verification into every spatial AI workflow. The AI surfaces information. Humans make the call.

When Will Spatial AI Be Widely Used in Construction?

Kalanithi puts the timeline at three to five years for major innovations. The foundations are already in place at companies like OpenSpace: large visual datasets, indoor location systems, and early progress tracking tools.

The broader shift is also underway. As more job sites capture visual data, the training datasets for spatial AI get better. Adoption of reality capture and the growth of spatial AI are not two separate trends. They are the same trend at different stages.

Frequently Asked Questions

What is spatial AI in construction? Spatial AI in construction means AI systems trained on visual and location data from real job sites. They help teams understand what is happening on a project and get the right information to the right person at the right time.

How is spatial AI different from large language models? Language models work with text. Spatial AI works with images, 360-degree captures, and location data. Construction is a physical industry, so spatial AI is a better fit than text-based AI for most job site problems.

What does OpenSpace's spatial AI do? It works at two levels. At the insights level it generates progress reports and schedule analysis from visual data. At the field level it powers agents that know where you are on a job site and surface relevant issues in real time.

When will spatial AI be common in construction? OpenSpace CEO Jeevan Kalanithi estimates three to five years for significant adoption. The core technology is already in place at leading companies.

What are the limits of spatial AI in construction? Reliability is the main constraint. Buildings are safety-critical and AI that is right 95% of the time is not good enough. Current tools are built to support human decisions, not replace them.

What is Auto Location? Auto Location is an OpenSpace technology that tracks where a smartphone is indoors without GPS. It connects every annotation and issue to a specific location on the job site.