How OpenSpace is building the data platform for a $13 trillion industry’s agentic future.
There’s a thesis gaining traction in technology circles that should concern every vertical SaaS company in construction: the rise of AI agents will eventually render most vertical SaaS products obsolete. Tools like Claude and ChatGPT are getting good enough that the text-driven workflows sitting on top of CRM databases—which is what most construction project management software really is—will be replicated and customized by agents that any company can configure for itself.
Your competition is no longer the other SaaS vendor in your space. Your competition is the frontier AI labs. And they are coming fast.
And I gotta say: this thesis feels mostly right.
So what wins? What is the durable position in a world where agents can replicate any text-based workflow?
The answer is data. Specifically, unique, proprietary data that agents need—but cannot generate themselves.
And that is the thesis of this piece: OpenSpace, the Visual Intelligence Platform for builders, is a data platform. And in the agentic era, the companies that generate the data layers for the $13 trillion construction industry will redefine how that industry can operate. Those companies that don’t… will be left out in the cold.
Two revolutions collide
There are two revolutions converging on construction right now. If you’re paying attention to only one, you’ll be blindsided by the other.
The first is Visual Intelligence: the shift from digital experiences centered around words, numbers, and spreadsheets to experiences that are centered on visual information, like maps, photos, and videos. This revolution was enabled by the explosion of camera technology: cheap, connected, powerful, everywhere. You saw this trend in social media (Facebook => Instagram), and Apple has embraced the term in the past couple years. People like pictures! And for real world industries like construction, Visual Intelligence is even more important, since these businesses are defined by what gets done in the field, not just what’s on paper. Visual Intelligence allows construction to shift from processes centered on documents and opinions, to managing execution with visual truth and facts.
The second revolution is Agentic AI: autonomous software agents that don’t just answer questions but take action. Agents that reason across data sources, compose multi-step workflows, and execute on behalf of users. Agentic AI is revolutionizing industries across the board. And agentic workflows for construction are maturing rapidly—not turnkey today, but getting there fast.
Here’s what the construction technology industry hasn’t fully grasped: these two revolutions are about to collide, and the companies that own the data layers where they meet will shape the industry’s next decade.
Why data beats analytics—every time
Let me be direct about where value is migrating.
In the AI era, document and text-oriented analytics will be commoditized. This is not speculation—it’s already happening. Any reasonably structured text-like dataset can be fed to an LLM, which can then generate insights, summaries, risk assessments, and recommendations. The analytical capabilities that SaaS companies spent years building into their products are being replicated by general-purpose AI in months. If your moat is “we built a nice dashboard on top of structured text data,” that moat is evaporating.
What cannot be commoditized is the data itself—especially data that is unique, difficult to capture, and essential for reasoning about physical reality.
Consider what the dominant construction technology platforms store: documents. Drawings, RFIs, submittals, daily logs, contracts, change orders. This is important information. But it is also exactly the kind of text-oriented, document-based data that LLMs are purpose-built to process. Feed a project’s RFI log and schedule into Claude, and it can generate a credible risk assessment today. Feed it the full document set and a well-designed prompt, and it can replicate significant chunks of what a project management platform does.
Now consider what OpenSpace captures: spatially indexed 360° imagery of active construction sites. Drone-captured aerial data. Smartphone photos with AI-generated indoor positioning. Voice-enriched Field Notes automatically tied to physical locations. Verified progress tracking data that combines computer vision with expert human review. All of it organized by space and time—a continuously updated, machine-readable record of physical reality.
No LLM can generate this data. No competitor can replicate it without doing what we’ve done: putting cameras on 90,000+ projects across 125 countries and building the Spatial AI to make sense of 60+ billion square feet of captured reality—operating at true scale to gather, organize, and make sense of this data.
This is the asymmetry that matters. While analytics gets commoditized, unique data capture becomes more valuable—because it becomes the necessary fuel that every agent, every analytics product, and every workflow must consume.
You know that phrase, data is the new oil? It’s true. Data is fuel. Analytics is the combustion engine, and today, you can build many engines (aka agents) for very cheap. But you can’t just fabricate the fuel from nothing.
And without fuel, you get nowhere, fast.
The OpenSpace data platform
When we unveiled the Visual Intelligence Platform at Waypoint 2025, we described it as the shift from document-centric to image-first workflows. We introduced AI Autolocation—a breakthrough that turns every smartphone into a real-time indoor positioning system, eliminating the need for costly and annoying hardware beacons. We launched OpenSpace Field, which lets any worker on a jobsite generate spatially indexed, AI-enriched data just by walking and talking. We acquired Disperse and launched Progress Tracking, which combines AI analysis with expert human verification to deliver trusted, visual progress data.
All of this drove real results—86% speed improvements in issue documentation for customers like Suffolk. 50% reduction in travel budgets. 41% fewer insurance claims for builders using the technology.
But those products are the face of our strategy. Here’s the heart.
OpenSpace is building the data platform for construction’s agentic future. Every image, every spatial index, every verified progress data point feeds a structured, queryable layer that any AI agent can consume. Our agents. Autodesk’s agents. Procore’s agents. Trimble’s agents. Oracle’s agents. Palantir’s agents. Anthropic’s agents. The data is the product.
This is the unlock. Not a one-size-fits-all application. A data-powered agent layer that adapts to each organization, each project, each trade.
I wish I could reveal more about this now, but rest assured, you’re going to hear a lot more from us soon.
Watch this space.
To the builders: this is for you
We started OpenSpace because construction deserves better tools—tools built around how the jobsite actually works, not how software companies imagine it works.
We know construction is one of the hardest things human beings do, and that no AI will replace a great superintendent’s judgment. What we can do is make that judgment better informed and faster to act on. Your business becomes more predictable: you find productivity problems at 10% completion instead of 50%. More profitable: fewer insurance claims. Fewer change orders. Accelerated payment cycles with progress verification based on visual truth. And most importantly, this technology works the way you work.
As one superintendent told me this year, what he loves about our tools is that “your new way is like the old way”—fast, visual, based on walking the job, seeing what’s real, making a call, and moving on. Not drowning field teams in (digital) paperwork.
The future of construction technology isn’t about making builders learn new software. It’s about making software disappear into the work.
To the technology platforms: this is for you
You’ve built companies that millions of professionals rely on. You orchestrate complex workflows across owners, GCs, subs, and architects—multi-party coordination that agents can’t replicate overnight. You’ve earned enterprise trust, contractual relationships, and real switching costs. These are meaningful advantages.
But think honestly about the timeline. Agentic capabilities are compounding. Construction companies are already experimenting with custom agents, and while these still require fine-tuning, they’re progressing fast. You have time to react—but not as much as you might think.
At the end of the day, construction runs on whether work got done, got done right, and got done on time. That’s how builders get paid. The data that answers those questions—objective, visual, spatially indexed, continuously updated—is reality data. Documents, drawings and models describe intent. Reality data describes what is. Every coordination workflow, every payment application, every design comparison ultimately bottoms out on that question: what actually happened on the jobsite?
We’re building the layer that makes every platform more intelligent. Agents built on Procore, Autodesk, Trimble, Oracle, or Palantir all get better when they can see—when they access a spatially indexed photographic record of what’s actually happening, not just what someone typed into a log. While you navigate the agent transition and compete with AI labs over who builds the best workflows, we provide the unique data every agent needs. That’s powerful for all of us, if we build it together.
To the technology community
OpenSpace sits at the intersection of two compelling theses. The vertical data thesis: the most durable AI companies will have proprietary, hard-to-replicate data in specific industries. The agentic infrastructure thesis: as agents proliferate, value accrues to the data and tool layers agents depend on, not to the agents themselves.
We have the data—the largest structured visual dataset in construction. The vertical AI expertise—Spatial AI, computer vision, indoor positioning, progress analytics with human-in-the-loop verification. And we’re building the agentic infrastructure that transforms our data into a platform.
Construction is a $13 trillion global industry that still manages its most critical asset—the physical jobsite—through documents and human memory. OpenSpace is building the data layer that replaces guesswork with ground truth. In the agentic era, that layer is not a feature. It is the foundation.
What comes next
When AI agents need to verify work completion, compare progress to schedule, or ground a recommendation in evidence—they’ll query OpenSpace data. Whether built by us, by Autodesk, by Palantir, or by a construction company using Claude—agents will need to see. We are how they see.
While SaaS companies take on the transition to an agent-native world, we’ll provide the fuel that every agent runs on. While the AI labs build increasingly powerful reasoning engines, we’ll provide the unique construction reality data those engines need to be useful. While construction companies begin experimenting with building their own agents, we’ll provide the framework that makes it possible.
The question isn’t whether the agentic era is coming for construction. It’s here. The question is whether you’re building on the data layer that makes it real.
We know where we stand. Come build with us!
