The rebrand came with little fanfare—GeoSpy, an AI tool that could pinpoint where a photo was taken by analyzing vegetation patterns and architectural details, became Raven in April 2026. On its face, a simple name change. Dig deeper, though, and the shift tells a different story: Graylark Technologies, the company behind the platform, was formalizing what law enforcement agencies had already discovered.
By the time Graylark announced Raven publicly, police departments in Miami-Dade and Los Angeles had purchased access to the technology. Documents surfaced through FOIA requests—first reported by 404 Media, with financials later detailed by Cybernews—showing both agencies had purchased access for approximately $85,500. The breakdown itself reveals something about who this was built for: two annual licenses at $5,000 each, 150,000 searches for $37,500, and a custom model tailored specifically to Miami-Dade for $38,000.
The custom model is the telling part. This wasn't off-the-shelf software being adapted for police work. This was purpose-built.
An OSINT Curiosity Finds Its Real Market
GeoSpy began as something of a curiosity in open-source intelligence circles. Founder Daniel Heinen, in an interview with Indicator last August, positioned it as a resource for journalists and researchers—people trying to verify the provenance of photos circulating online, perhaps, or investigating claims made in social media posts. The technology was genuinely novel: feed it an image, no metadata required, and the AI would infer location from visual cues most humans wouldn't consciously register. Soil composition. Tree species. The particular way shadows fell across a building facade.
But the product's trajectory diverged sharply from that origin story. When Graylark launched Raven in a May press release, the capabilities had expanded far beyond simple geolocation. The platform now advertised "meter-level precision" for street-level targeting. It could identify vehicles. Analyze entire scenes. A beta deepfake detector was in the works. Perhaps most significantly for law enforcement applications, it could now process low-quality CCTV footage, screenshots, even night video—exactly the kind of degraded imagery that turns up in investigative work.
These weren't features aimed at journalists on a deadline.
The Access Question
Visit Raven's website and the positioning is explicit: the platform is "available for purchase to qualified law enforcement agencies, enterprise users, and government entities." No public trial. No consumer tier, no indie researcher tier. The deliberate narrowing from GeoSpy's earlier openness to the OSINT community is complete.
The product architecture reinforces this. Raven includes case management features—shared workspaces, multi-source evidence handling, collaborative pinning—built for team-based investigations rather than individual analysis. It's a workflow tool as much as it is a piece of AI. Oceanir, a competitor positioning itself as a "Raven alternative," has noted in its own marketing materials that Graylark shifted to law enforcement and enterprise-only access after privacy concerns about the technology began circulating in tech media outlets in early 2025.
Whether that timing was reactive or strategic, Graylark hasn't said.
Money Trails and Opaque Signals
For a company that's sold its technology to multiple police departments, Graylark's funding picture remains surprisingly murky. A third-party startup directory lists $73,900 in pre-seed funding from Hatch and AI Grant, last updated in March, though that figure hasn't been corroborated through SEC filings or primary sources. No Form D appears under Graylark Technologies, GeoSpy, or obvious related entities. The company hasn't issued a funding announcement. General Catalyst—a venture firm that raised $8 billion last October—hasn't disclosed any investment, despite speculation in some corners of the startup community.
That opacity is striking given how public the company's pivot has become. Graylark's corporate website now frames its AI as enabling "mission success," with explicit appeals to law enforcement, government agencies, and enterprise buyers. The language is careful, sanitized—but the intent is clear.
The transformation from OSINT novelty to restricted-access government technology is now complete, formalized not just in a name change but in product features, pricing models, and sales positioning. Whether Graylark sought out law enforcement as customers or law enforcement found Graylark first remains an open question. What's certain is that both sides found what they were looking for.
And the technology, whatever its original purpose, has found its market.
