Saab launches AI agents for future aircraft and autonomous flight

New AI agents for combat

Saab has officially unveiled a suite of in-house AI agents designed to support future aircraft platforms. These agents are built to master tactical decision-making in complex operational environments where speed and accuracy are critical. The technology is intended to function as a teammate for pilots inside a cockpit or as the primary brain for uncrewed air vehicles.

The company recently completed testing these new technologies on the Gripen E, which is a multirole fighter jet. These tests demonstrated the firm's ability to develop, train, and fly its own artificial intelligence for air combat operations. By integrating these agents, Saab aims to provide a software-driven advantage in modern warfare.

The primary focus of these early tests involved Beyond Visual Range (BVR) combat conditions. BVR refers to combat situations where aircraft engage targets at distances too great for the human eye to see. This type of engagement requires rapid computer processing to manage multiple variables simultaneously.

The AI Pilot Academy

To support these new agents, Saab is launching the AI Fighter Jet Pilot Academy. This facility is designed to do more than just test the software; it provides the infrastructure needed to create, train, and evaluate the AI systems. The goal is to move beyond simple success in a single flight and toward a dependable capability.

Peter Nilsson, the head of Advanced Programs at Saab’s aeronautics unit, noted that the true purpose of the academy is to ensure the AI can learn and improve over time. He emphasized that the strategic advantage of AI-driven data comes from reducing the distance between experience and field-led capability. The academy serves as the bridge between raw data and usable combat experience.

This infrastructure is necessary because AI requires a structured environment to understand the nuances of flight and combat. By establishing a dedicated academy, Saab can systematically refine how these agents interpret complex signals. This ensures that the AI remains relevant as new threats and technologies emerge on the battlefield.

Data sovereignty and Sferical AI

Saab is also becoming a partial founder in a Swedish government-backed initiative called Sferical AI. This move is intended to scale the computer capacity needed to run the AI agents without relying on external commercial giants. Nilsson stated that without this initiative, the company would have to pay Google or Amazon for computing power.

By using a domestic initiative, Saab intends to maintain full control over its sensitive data. Relying on public cloud providers can create risks regarding who has access to proprietary defense information. Keeping this capability within Sweden is described by the company as having enormous value for national security.

Sferical AI utilizes Nvidia’s DGX SuperPOD architecture to develop Large Language Models and Vision-Language Models. These models are used to facilitate communication between the AI agents and human operators. This setup allows for a more seamless interaction between the software and the people directing it.

Scaling flight experience

One of the most significant aspects of the Sferical AI infrastructure is its ability to compress time. When applied to a one-hour simulated training flight, the system can produce the equivalent of one billion flight hours for a human pilot. This exponential computational curve allows the AI to

Future aircraft programs

While the current tests focus on the Gripen E, Saab plans to leverage this in-house AI technology for its future air programs. This includes the development of a trio of unmanned aircraft known as the A1, A2, and A3. These projects are part of a larger path toward designing a new loyal wingman drone.

A loyal wingman is an uncrewed aircraft designed to fly alongside and support a manned fighter jet. By using the AI agents developed for the Gripen, Saab can ensure these drones are capable of autonomous flight. This integration will be key to the next generation of aerial warfare.

The company's move reflects a broader trend toward software-defined defense. By mastering the ability to train and fly their own AI, Saab is positioning itself to deliver autonomous systems that can adapt to changing environments. The transition from human-led to AI-assisted combat is a central pillar of their long-term strategy.