Dataminr CEO Ted Bailey confirmed that First Alert Advanced will go live on NATO's Maven Smart System this week, days after a $318M five-year DoD contract was announced. The deal provides Dataminr's capabilities—situational awareness, force protection, and indications and warnings—to up to 3 million licensed users across all DoD commands and field activities. NATO is now Dataminr's second-largest customer.

First Alert Advanced uses what Dataminr calls "multimodal fusion"—scraping internet sources, AIS/ADS-B transponders, imagery, video, audio, and 150 languages—to generate near real-time intelligence. The system includes autonomous "intel agents" that scan, synthesize, and brief users, plus a "tactical predictive intelligence" module that uses historical patterns to forecast events. Bailey claims the platform detected the first US-Israeli strikes on Iran 15 minutes before mainstream media and three minutes before CENTCOM's official operation start time.

Integration with NATO's Maven Smart System—Palantir's AI platform connecting disparate C2 systems—allows Dataminr to fuse open source intelligence with military sensor data. This embedding enables cross-domain synthesis: publicly available information corroborated and enriched by classified feeds. The architecture positions Dataminr as infrastructure rather than a standalone tool.

The Pentagon and Intelligence Community have spent years racing to operationalize the explosion of open source data. Dataminr's win reflects a shift toward treating OSINT fusion as a baseline capability rather than a niche function. The company is expanding beyond the US—17 other NATO militaries across 15 member nations are either deploying or evaluating the platform.

The predictive intelligence module raises operational and doctrinal questions. After the Iran strikes, the system reportedly forecast Strait of Hormuz disruptions, oil price spikes, and high-value target deaths. Bailey attributes this to LLMs trained on historical event patterns. How commanders weight AI-generated forecasts against human analysis—and where accountability sits when predictions miss—remains an open doctrinal problem.