Using AI to take a manoeuvrist approach to defence legacy
In defence, advantage comes from applying force where it matters most. AI is now making it possible to understand complex legacy systems well enough to focus effort where it delivers the greatest effect.
UK defence’s digital landscape is a deeply interconnected system of systems, spanning C4I platforms, communications infrastructure, digital services, and operational technologies across the MoD and its wider ecosystem. Many of these systems are decades old, tightly coupled, and only partially understood.
Legacy as a structural feature, not a failure
This creates a structural challenge: obsolescence is not an anomaly, but an inherent feature of defence at scale. Efforts to modernise have historically struggled not due to lack of intent, but because the problem itself is difficult to define. When systems are tightly coupled, changing one affects many others, making understanding as complex as transformation.
The result is a persistent cycle of legacy, limited visibility, and risk aversion, where the safest decision is often to leave systems unchanged.
The cost of working around legacy
Limited visibility into system dependencies makes it difficult to identify vulnerabilities or respond quickly to incidents. Workarounds inevitably emerge, created by users seeking faster access to functionality or developers adapting systems beyond their original design.
While often effective in the short term, these workarounds tend to be poorly documented, difficult to support, and frequently operate outside formal control frameworks. Historically seen as shadow IT, and now increasingly as shadow AI, they can demonstrate what is possible but rarely scale into core environments due to security requirements, integration challenges, and the need to meet strict approval and governance standards. Instead, they introduce new, ungoverned dependencies into an already opaque ecosystem, adding to complexity and creating the conditions for the next generation of legacy.
Legacy systems are not only a constraint on modernisation, they are also high-risk targets. Hidden dependencies and fragmented integrations create vulnerabilities that are increasingly easy to exploit as AI capabilities advance (like Anthfropic’s latest model Mythos demonstrates). Improving transparency and control through modernisation significantly reduces this risk.
System understanding through AI
This is where artificial intelligence presents a different kind of opportunity. A new class of agentic AI systems can interpret and reason over software estates, reconstructing system understanding directly from artefacts such as code, data structures, and runtime behaviour – a practice we call AI-driven legacy archaeology.
This makes it possible to surface operational logic, workflows, and dependencies that would be extremely difficult to identify through manual analysis alone, shifting modernisation from assumption to evidence.
Achieving this depends on a structured approach. Systems need to be broken down into manageable components, with clear decisions about what to examine and what needs to be achieved. This is where process design and expertise come into play, ensuring AI is applied in a way that produces meaningful and usable insight in complex environments.
With that structure in place, AI-driven analysis becomes significantly more effective, enabling organisations to extract reliable understanding from systems that would otherwise remain difficult to interpret.
Focus effort where it delivers the greatest impact
Applied to legacy modernisation, this enables a more manoeuvrist approach. Instead of large-scale transformation programmes, organisations can use AI-driven insight to focus effort where it delivers the greatest impact, based on a clearer understanding of system interactions.
This improves the ability to integrate new technologies and enables faster adoption of capabilities such as AI, autonomy, and advanced sensing.
In practice, this means targeting specific interventions – improving interoperability, reducing reliance on manual workarounds, or addressing risk in tightly interconnected systems. Here, AI is not replacing systems, but enabling more precise and confident change.
AI acts as a force multiplier, improving decision-making, resilience, and the effective use of resources.
Breaking the cycle of legacy
AI-driven legacy modernisation offers a path to make systems visible and move towards incremental, controlled change.
The opportunity is not simply to modernise legacy systems, but to break the cycle in which complexity limits change. With the right application of AI, defence organisations can move from reacting to legacy constraints to actively shaping more secure and adaptable systems.
