Archer unveils Zee, aviation-specific AI model for flight safety

On July 15, 2026, Archer Aviation Inc. (NYSE: ACHR) announced Zee, which the company describes as the world's leading aviation-specific foundation model. Unlike general-purpose AI systems, Zee is purpose-built for aviation and delivers what Archer calls a unified aviation intelligence platform. The model integrates multiple data streams: ADS-B signals (automatic dependent surveillance-broadcast, a real-time aircraft tracking system), air traffic control communications, navigational maps and charts, aircraft state data, terrain information, and weather data.

Zee is trained on real-world operational data collected through Archer's proprietary data pipeline and aggregated from a global network of over 6,000 ADS-B receivers. This distributed sensor network gives Archer access to continuous flight data that Zee can learn from and refine its models against. The foundation model is architected to operate in two modes: offline and on-device (requiring no connectivity) or as a server-hosted solution, a dual-mode design that Archer says is critical for diverse aviation use cases ranging from air taxis and unmanned aerial vehicles (UAVs) to commercial airliners and air traffic management systems.

Archer's AI effort is substantial. The company has assembled a team of nearly 100 researchers and engineers led by Mario Srouji, who joined Archer last year after working at Apple. The effort is advised by Professor Ruslan Salakhutdinov, who previously served as VP of AI Research at Meta and Director of AI Research at Apple, lending deep expertise from two leading AI research organizations. In the announcement, Adam Goldstein, Archer's founder and CEO, framed the strategic vision: "We are building an intelligence layer for the entire aviation system with Zee. The company that owns the data and the foundation model will help lead the aviation industry into the next era of flight."

The scale of the opportunity Zee addresses is substantial. On a typical day, more than 45,000 flights move through American airspace alone, each generating a constant stream of radio calls, navigation inputs, and aircraft state information. Pilots and air traffic controllers currently synthesize these disparate signals in real time. Zee's unified approach aims to deliver lower latency and better performance while enabling on-device operation that requires no connectivity—critical for safety-critical aviation environments where connectivity cannot be assumed or relied upon. Archer is currently in discussions to deploy Zee through pilot programs with governments, airlines, and other industry partners, with initial applications focused on airline operations, airspace management, and copilot assistance, all aimed at improving flight safety and efficiency.

Archer's announcement of Zee reflects a deliberate strategy to build proprietary AI tailored to a specific operational domain—aviation—rather than relying on general-purpose large language models. The model's architecture, combining data from air traffic control communications, aircraft state, and environmental sensors into a single training framework, addresses a real constraint in aviation: pilots and controllers currently piece together disparate data sources in real time. By unifying these streams, Zee aims to reduce latency and enable decision-making on-device without connectivity—a critical capability for safety-critical systems where network outages or delays cannot be tolerated.

The team's leadership matters here: Mario Srouji's background at Apple and Ruslan Salakhutdinov's dual tenure as VP of AI Research at Meta and Director at Apple signal that Archer is staffing this effort with researchers from two of the most advanced AI organizations. With nearly 100 researchers and engineers on the AI team, Archer is investing substantially in the model's development and deployment.

The 6,000 ADS-B receivers constitute a significant data moat—the ability to collect ground truth from a globally distributed sensor network gives Archer a training advantage that competitors without such infrastructure cannot easily replicate. The mention of more than 45,000 flights daily in American airspace alone underscores the volume of data flowing through this system, which Zee is designed to process as a coherent whole rather than as isolated signals.