SATIM announces a new product that enables ISR programmes with end-to-end control of their AI capabilities, allowing for tailoring the AI models to sovereign data, defence priorities, and evolving operational needs.
Kielce, Poland – September 8, 2026 – Today, SATIM has announced achieving a breakthrough in developing sovereign AI solutions for SAR data exploitation: OREC Evolve marks a new direction for the OREC SAR intelligence portfolio.**
OREC Evolve is a response to the needs of intelligence end-users who see AI as the natural solution to processing and interpreting large amounts of data, but are unable to share the data to 3rd parties for training. OREC Evolve changes this paradigm by bringing the expertise behind OREC directly to the user, enabling them to train and adapt proven SAR AI models on their own data through an optimized, SATIM-developed approach.
“AI model adaptability is what makes OREC Evolve a perfect fit for sovereign ISR programs. The challenge was to shorten the AI training and adaptation process and reduce its complexity for the user as much as possible. This way, organizations can build effective SAR AI capabilities without us knowing what SAR data they are using OREC on or what their mission priorities are. I'm proud that our team has once again shown why we're at the forefront of SAR data exploitation." – Stanisława Porzycka-Strzelczyk, CTO of SATIM
Capability from day one, adaptability along the mission
From the moment OREC Evolve is installed, end-users will get the AI capabilities for automated object detection and classification on SAR data, with leading performance, support for on-premise and airgapped deployments, and C2 and C4ISR system integrations.
OREC’s sensor-agnostic architecture supports multiple providers, such as Capella Space, ICEYE, Synspective, Umbra, or Airbus, and proved to be compatible with sensors it hasn’t been trained on. In such cases the performance can be improved by fine-tuning.
OREC Evolve's adaptive AI lets users fine-tune OREC's models against the sovereign imagery, sensors, and object-recognition priorities as missions evolve. What’s most important is that it doesn’t require AI experts and software engineers to build an effective automated SAR data analysis capability in a sovereign ISR program.
OREC Evolve will help users:
- – automate dataset labelling in user-defined areas,
- – simplify data preprocessing,
- – perform quality assurance and compare model candidates,
- – introduce new object signatures into datasets,
- – and manage model adaptation entirely within their own environment.
The goal is to allow ISR organizations to develop and maintain highly specialized SAR AI capabilities without having to build an entire SAR machine-learning stack from scratch.
“We have been involved with SAR and machine learning since 2012. Working on automating exploitation of such complex data since that early has given us years of advantage which we now leverage to let NATO and allied end-users stay ahead of their adversaries and maintain the edge for years. OREC Evolve is a showcase of the innovation at the heart of SATIM.” – Jacek Strzelczyk, CEO of SATIM
The effect is two-fold: on one side, sovereign ISR programs obtain the AI analysis capability quickly and secure access to further model updates and improvements provided by SATIM. On the other hand, they further boost model performance, specifically for their sensors, use cases, and priorities. The AI performance raises to the level not available with other approaches and coverage stays relevant as the situation changes over months and years.
A product announcement during MSPO 2026
OREC Evolve is now entering the first phase of operational user trials and will be further developed in the directions identified via user feedback. Program managers building or scaling sovereign ISR capability can book time with SATIM at the show or reach out directly to discuss entering some of the remaining trial positions.
The product was announced first at the MSPO 2026, the 34th International Defence Industry Exhibition in Kielce, Poland.
