Autonomous advantage: How SitaWare is advancing agentic AI

From data challenges to domain expertise, Systematic is building the agentic AI capabilities that tomorrow's militaries will depend on

Agentic AI has strong potential for militaries, boosting capabilities in decision-making, communications and beyond. Although the technology is still in the early phases of development, its rapid evolution means militaries and industries must work closely together to seize the opportunities ahead.

AI algorithms are already a core part of the SitaWare suite, from the AI Assisted Toolbox designed to improve work processes, to the use of machine learning for anomaly detection in the maritime domain.

In particular, SitaWare Insight provides AI-powered analysis for a range of tasks, including managing the full intelligence cycle and empowering commanders across the battlespace. New features enable operators to partner with a large language model (LLM) chatbot as a ‘wingman’ for mission support, among other capabilities.

But agentic AI would represent a major leap forward and is now a key focus for SitaWare.

What is agentic AI?

“The AI systems we know today operate in a relatively passive manner. They are designed to conduct bespoke tasks based on individual user prompts,” explained Henrik Sommer, Director at Systematic and former Brigadier General in the Royal Danish Army.

“But with agentic AI, multiple AI agents would collaborate to accomplish complex tasks independently,” Sommer continued. “While there would still be a ‘person in the loop’, the technology could function with only minimal human interaction if required.”

Such autonomous capabilities would present huge advantages for militaries, quickly and effectively analyzing and communicating data in such areas as terrain, weather and enemy analysis.

For Systematic, this means building connections across a range of individual AI agents to foster an agentic AI capability. According to Sommer, this work is already well underway. Based on feedback from military operators, the first agentic AI capabilities are expected to be integrated into SitaWare products in early 2027.

Play video

The data challenge

Like any rapidly evolving technology, agentic AI presents a series of challenges, specifically the need for credible data.

“If you don’t have the data, don’t talk about AI – it doesn’t matter,” warned Sommer. “SitaWare has a strong advantage here due to the established infrastructure of our suite of software.”

The SitaWare suite of products enables operators to effectively communicate from the strategic to the tactical level and incorporate inputs from more than 50 nations.

“SitaWare is unique because it has been used at the edge, but also at the highest joint headquarters,” Sommer said. “The whole family is closely knitted together: it can get the data, it can process the data, and it can push the data forward.”

Systematic has been experimenting with the technology for some time, Sommer noted. This will determine whether the agents are coordinating together effectively and consider whether the output can be trusted – and where do the users want to have “the person in loop” to validate the AI generated products.

“Every time you work with AI, you must assess the outcome. Is it trustworthy? That's why we need to experiment with it, because there are risks with AI,” stated Sommer, pointing to such dangers as AI hallucinations. AI hallucinations are confident and plausible sounding outputs done by AI that prove to be false or misleading and generally occurs when systems produce information not justified by their training data.

SitaWare software is designed with resilience in mind, no matter the connectivity available, a crucial consideration for militaries in the multi-domain battlespace. SitaWare is built to operate in such environments.

“If you are dependent on high-bandwidth connectivity, more often than not you will lose it,” explained Sommer. “SitaWare software mitigates this risk through SitaWare Tactical Communication, which can transmit data in low-speed connections and in a disconnected, denied, intermittent, or low-bandwidth (DDIL) environment.”

Domain knowledge

Sommer stressed that it is vital to engage with software companies that bring the right IT and domain knowledge. In the military context, software companies must develop agentic AI that can effectively operate in the multi-domain battlespace and mitigate by design the risks.

“C4ISR software is so complicated, particularly when it comes to interoperability,” said Sommer. “It is hard to achieve that interoperability across domains, because they are often still ‘stove-piped. You must have the right expertise and understand the numerous NATO Standardization Agreements to solve the interoperability challenge as we have done with SitaWare.”

The rapid pace of change in the AI market means it is essential to embrace a range of innovations. Systematic has developed its own AI algorithms, including a maritime anomaly detection agent. This was developed in-house, but trained on Danish maritime data, so it represents a mix of in-house expertise and military experience.

However, it is also possible to attach third-party algorithms to SitaWare, including AI models and tools developed by customers. Sommer explained that the software provides users with the capability to test and integrate new algorithms. An important capability because of the ever-increasing speed of development of new AI models.

User cooperation

AI advances are built on close cooperation with customers. Crucially, Systematic does not simply hand customers the software tools, it demonstrates the capabilities. This collaboration is essential when it comes to a rapidly evolving technology like AI.

For example, Systematic works closely with military operators through the ‘Battlelab’ concept, which involves Systematic experts collaborating with customers to test the software and teaching them the full capabilities of agentic AI.

Sommer stressed that such consultancy support continues on an ongoing basis long after operators have acquired the technology.

“We don't just deliver the software and say, ‘OK, we have trained some of your guys, now you can go do it by yourself,’” explained Sommer. “Communication between the user and consultants from Systematic is extremely important to effectively use the products and reach their potential.”

Future focus

As new use cases emerge for agentic AI and militaries demand new workflows, Sommer noted that cooperation between military customers and industry specialists like Systematic will be vital in the years ahead.

“There could be 10,000 different ways that you can use agentic AI,” Sommer noted. “We are helping militaries to understand the potential of our product - that is not just doing presentations, it is going out and engaging with the customer.”

“It’s all about the triangle between technology, doctrine and organization – that applies in an agentic AI world just as it does with any new innovation,” said Sommer.