Saab Introduces New AI Agent Suite for Next‑Generation Aircraft

BELFAST — Saab, the Swedish aerospace firm, has introduced an in‑house family of AI agents intended for future aircraft platforms, after conducting trials on a Gripen E multirole fighter.

The move demonstrates the firm’s ability to develop, train, and fly its own AI for air‑combat operations, according to a company statement released today.

The agents employ various AI techniques to master tactical decision‑making in complex operational environments, the company added. They can operate either alongside a pilot in the cockpit or autonomously from an unmanned air vehicle.

Central to the initiative is Saab’s AI Fighter Jet Pilot Academy, which is dedicated to developing the agents while building the infrastructure needed to create, train, and evaluate them, the statement noted.

The challenge now, according to Peter Nilsson, head of Advanced Programs at Saab’s aeronautics unit, is “how we turn this success of flying with the agents into a dependable capability that can learn, improve, and remain relevant over time.” He noted that this is the true purpose of the AI Fighter Pilot Academy, in comments shared with Breaking Defense and other media ahead of the launch.

Nilsson emphasized that the strategic advantage of generating AI‑driven data increasingly stems from narrowing the gap between experience, learning, and field‑led capability.

Having a software‑driven platform like Gripen that can keep pace with this processing speed is of enormous value, Nilsson added.

The AI agents have so far been tested primarily in Beyond Visual Range (BVR) combat scenarios, echoing a 2025 trial in which Helsing’s Centaur AI took control of a Gripen and executed BVR maneuvers autonomously. Looking ahead, Saab intends to apply its proprietary AI to future air‑program development, including a trio of unmanned platforms—A1, A2, and A3—that are being designed as part of a new loyal‑wingman drone concept, as previously reported.

To scale the computational resources needed for its AI agents, Saab has become a partial founder of a Swedish government‑backed initiative called Sferical AI.

Without Sferical AI, Nilsson noted, “you would have to go to Google or Amazon … and pay some money,” which could mean “you maybe don’t have full control of your data.”

Retaining such capability domestically is “enormous” for Saab, he added, pointing out that the initiative was officially launched at a ceremony in Stockholm the previous month.

According to a Saab slide presented by Nilsson, Sferical AI leverages Nvidia’s DGX SuperPOD architecture to enable the development of Large Language Models and Vision‑Language Models that support agent‑operator communications.

Applying Sferical AI to a one‑hour simulated training flight can generate the equivalent of one billion pilot flight hours. “The exponential computational curve is … really hard to comprehend,” Nilsson explained.

Source link