The geospatial industry has never been short on capability. That said, industry leaders have learned to look beyond the technology, to lead with problems instead of solutions. The most successful seek to find the places where people are improvising, stitching together workarounds, doing something difficult and expensive because no better option exists, and to start there.
It’s a core philosophy of Philadelphia-based Exyn Technologies, a leading developer of autonomous LiDAR-based mapping systems. The company builds systems that navigate and capture data in GNSS-denied environments—underground mines, bridge substructures, confined industrial spaces—where conventional positioning solutions fail and traditional survey methods can be dangerous, slow, or impossible. When COO Ben Williams joined, the company was about five years old and had spent much of that time developing its core capability: SLAM-based autonomous navigation for aerial systems operating in GNSS denied or limited areas. Although the technology performed exceptionally well, it took a chance customer encounter to reveal where the real value lay.
A potential buyer walked past a demo, watched the drone navigate autonomously, expressed polite interest, then pointed at the point cloud the system was generating as a byproduct of figuring out where it was, and said he wanted that.
It was a small moment with large implications. The autonomy, it turned out, was the means. The data was the end. And the overall solution needed to be organized around delivering the data, with the autonomy as the mechanism that made capture possible in places no other approach could reach.
That realization hardened into a philosophy—and a way of finding the next problem worth solving. “Basically, we look for the duct tape,” Williams explains. “We find those early adopters who have a problem that they’re trying to solve with improvised solutions.”
It became the engine behind everything Exyn has built since and the implications produced something more valuable than a better tagline: a self‑reinforcing loop in which better autonomous navigation produces richer, more complete data, and richer data enables more confident autonomous navigation.
Inside Exyn, that loop sits at the heart of a broader flywheel: each improvement in reliability and data quality makes it easier to win and serve the next customer — and to carry the same playbook into the next market.
Forging the Flywheel
The first market where Exyn really applied its duct‑tape logic was underground hard rock mining, specifically the process of tracking the geometry of mined-out voids to inform blast planning and structural safety assessment. It is a demanding environment: poor lighting, irregular geometry, significant safety concerns, no GNSS and a workforce accustomed to improvising with whatever tools are available.
It is also, Williams readily acknowledges, an extremely narrow customer base. Exyn’s longest‑running customer relationship with gold miner Dundee Precious Metals has become as much an education in industrial workflows as a source of revenue.
He says, “They were very forward-looking in terms of new technologies and data-centric operations. But they were also a key reason for our strategic shift.”
Understanding how Dundee operated meant going well past the surface problem. It meant learning how a survey team descends into a mine, captures data, geo-references it, and processes results, ideally before surfacing, so that the next shift already has the information it needs before anyone goes back underground.
It meant understanding how buyers think about risk, and how to make an investment in unfamiliar technology feel like a solved problem rather than an experiment. It meant learning, as Williams puts it, to go in and say “I know I’m wrong here—tell me where” as a way of disarming defensiveness and getting to the real operational realities faster.
For Exyn, mining took four years to develop into a reliable, repeatable business. More importantly, it shaped the new playbook for the firm. Exyn now understood how to find the right early adopters, how to translate workflow insights into product features, how to build a go-to-market approach that could be documented, staffed against, and repeated.
A Horizontal Pivot
The transition into geospatial took roughly half the time mining had required. Some of that compression was experience; some of it was a conceptual adjustment that turned out to be as important as any technical advance.
Exyn’s initial instinct was to approach adjacent markets the way most companies do: as verticals. Construction was a vertical. Infrastructure was a vertical. Each had its own buyer profile, budget cycles, and procurement processes. The problem was that none of them had a clean line item for “autonomous mapping system” and the teams who most wanted the technology were often not the ones with authority to buy it.
“Everyone we talked to was like, ‘This is great, we definitely need this. Someone should definitely buy it,'” Williams says.
The insight that unlocked the market was reconceiving it horizontally rather than vertically. The teams actually using Exyn’s systems were not construction teams or infrastructure teams or mining teams in any rigid sense—they were data collection groups, operating wherever reliable spatial data needed to be captured: construction sites, bridge decks, maritime environments, confined industrial spaces, post-event assessments. The client’s industry was less important than the data flow. And the underlying workflow, stripped of context, looked remarkably consistent across all of them.
“Our focus really is on geospatial groups, those people who are doing the work of capturing data wherever it’s needed,” Williams says. “That’s how we have a broader portfolio of verticals, but we’re trying to serve the need of the surveyor or the data collection group in all of those areas.”
That shift in focus has quietly reshaped the kinds of partnerships Exyn pursues.
The Seafloor Setup
One of the clearest expressions of that horizontal strategy—and of Exyn’s broader platform philosophy—is its partnership with Seafloor Systems, a Sacramento-based developer of unmanned surface vehicles for hydrographic survey. The collaboration tackles a persistent hard problem in infrastructure inspection: what happens to a survey when it goes underwater, under a bridge, or into any environment where GNSS disappears and the tools hydrographic surveyors have relied on for decades simply run out of answers.
One in four bridges in the United States is 75 years old or older. Many of them have never been surveyed below the waterline with anything approaching the accuracy available above it. The tools exist to complete the task in the way of multi-beam echo sounders, robotic total stations, inertial navigation, etc. But they break down in the confined, complex environments where the most critical structural questions arise. Dense pile fields defeat line-of-sight positioning. Low overhead clearance eliminates conventional vessel access. The areas that most need inspection are often the areas hardest to reach.
Seafloor had been building unmanned surface platforms since around 2010, originally for marine construction. Exyn brought SLAM. Together, they developed a workflow in which a compact unmanned surface vehicle outfitted with both a multi-beam echo sounder and an Exyn SLAM LiDAR sensor surveys the underwater environment while the SLAM system simultaneously maps the above-water structure and generates the positioning trajectory that the sonar data requires.
The technical elegance of the solution sidesteps the GNSS problem entirely. The SLAM point cloud is registered to a geo-referenced control point cloud captured beforehand with static terrestrial scans. The transformation matrix from that registration is then applied to the SLAM-derived trajectory, converting it from an arbitrary coordinate system into a fully geo-referenced one. That trajectory feeds into standard hydrographic processing software as if it came from a conventional GNSS-inertial navigation system.
For Exyn, the partnership also illustrates the platform logic Williams describes: rather than building a full vertical solution—hardware, processing software, reporting, asset management, client delivery—Exyn exposes its core capabilities through APIs and integration pathways, enables partners like Seafloor to build on top of them, and concentrates its own energy on making the underlying system faster, more reliable, and easier to deploy.
“We always go in with software and data partners,” Williams says. “We don’t think we’re so smart that we can solve all that stuff. Why try to displace people that are already doing great work? Let’s enable them to be more successful—and it keeps the flywheel spinning.”
Richer Data, Smarter Autonomy
The flywheel concept is easy to describe and genuinely hard to build. The companies that fail to build it, Williams suggests, are usually not failing on the technology. They’re failing to go deep enough, early enough, in any single market to build the understanding that transfers.
Exyn’s rule is one new market entry at a time, never more. The process does not move to the next market until the current one is understood well enough to hire and train people against it, to institutionalize it, as Williams puts it, so that the knowledge is in the organization rather than in a few individuals’ heads.
“A startup takes a bunch of smart people, they try to figure some stuff out, and then it’s in their head. Then they document it, then they institutionalize it, and then you rinse and repeat,” he says. “That, I think, is how a company grows up.”
The same discipline extends to how Exyn thinks about AI, a topic that has become unavoidable in any geospatial conversation. Williams draws a careful distinction between what he calls LLM-type AI and the deterministic, algorithmic AI that actually runs Exyn’s systems. The core navigation and mapping are deterministic by design: consistent, trackable, auditable, and eventually, he hopes, applications certifiable for safety-of-flight.
Classification models that sit above the core system can be non-deterministic, because that’s where the flexibility pays off. Distinguishing dust from a hanging wire—one of the cases Exyn explicitly designs for—or flagging a change in tunnel profile or vegetation encroachment on a power corridor are problems where trained models outperform hand-coded rules. But the boundary between the deterministic core and the probabilistic layer is drawn deliberately, not by default.
“For a tool like ours, we want deterministic solutions for the core autonomous navigation and the core mapping,” Williams says. “What gets interesting is on the classification layer on top of that.”
The Next Frontier
The biggest growth opportunity Exyn sees right now is in the broader geospatial market.
But the use cases he is most interested in are not the ones that already exist. They are the ones that only become possible when the cost and friction of capture drop far enough that teams who previously could not justify scanning suddenly can. The zero-to-one opportunity, as he frames it: not doing something more efficiently, but doing something that was not being done at all.
“If it’s too expensive and too inconvenient for someone to do something, they just won’t do it. It’s not that they’ll do it less,” Williams says. “If you then have a low-cost, high-speed scanner, all of a sudden they can go from zero to one. And it opens up whole new categories that just couldn’t exist before.”
The companies best positioned to capture that opportunity, in his view, are the ones that start from problems rather than products, that find the places where people are already improvising, already building workarounds, already duct-taping one system to another because no good solution exists yet. Those are the signals worth following. Those are the markets worth going deep on.
The geospatial industry has spent decades getting extraordinarily good at capturing the world. What comes next is about what happens after the capture—who acts on the data, what decisions change, and which problems finally get solved. For the companies that have learned to ask those questions first, the flywheel is already spinning.