Terma operates under conditions that make AI adoption particularly difficult. Strict compliance requirements, export licensing obligations, and data security constraints are not peripheral concerns — they are built into every process. At the same time, growing demand for defense systems and a tight talent market create real pressure to find efficiency gains without simply hiring more people.

The IT department had identified AI as a strategic priority, but encountered a familiar barrier. There were no active AI transformation projects in the business units, no shared roadmap, and no tools employees could relate to their daily work.

“The biggest challenge right now is finding someone who has the time to make something work — and document the benefit,” the IT department noted. “We’re collecting use cases hoping it will be enough to get someone to approve the resources.”

The project applied and reconstructed the 4P Hybrid Intelligence Maturity Framework (Sherson et al., 2025) — spanning Projects, Platform, Policies, and People — to map Terma’s current AI readiness across two DSR iterations.

In the first iteration, Mikkel presented a GPT-based artifact to a focus group of service managers, using fictional radar product data to simulate a real workflow: retrieving parts, checking lead times, and cross-referencing export license requirements — a process that normally requires manual ERP lookups for each component. The demonstration produced genuine curiosity. It also produced a precise finding: Terma scored between level 1 and 1.5 across all four pillars. The People dimension was not resistance — it was readiness without direction.

The focus group also revealed a structural limitation in the model itself. Three maturity levels were too coarse: employees fell between levels and needed finer gradations to locate themselves meaningfully. This feedback directly shaped the second iteration, in which Mikkel reconstructed the 4P model with four levels and domain-specific language drawn from the defense context — including compliance layers, export control, and distinctions between different employee groups including specialists and those operating outside formal processes with “shadow IT.”

This co-created reconstruction is itself a Hybrid Intelligence outcome: the model became more useful precisely because the people who would use it helped shape it (Sherson et al., 2025).

The project established a baseline for AI maturity at Terma that did not previously exist — and demonstrated that building organizational support for AI transformation requires more than deploying tools. It requires a shared language, a relatable framework, and the involvement of employees in defining what maturity actually looks like in their context.

The adapted 4P model now functions as a leadership artifact: a structured way to identify where Terma stands, where it needs to move, and what the gap between current and required maturity looks like for any given AI initiative. Visualized as a radar chart, it makes strategic decisions about AI investment more legible — and more defensible.For the defense sector more broadly, the case illustrates that AI transformation in regulated, security-sensitive industries follows the same human logic as anywhere else: people must understand what is being asked of them, see the value in it, and feel genuinely involved in shaping it. The 4P framework (Sherson et al., 2025) offers a rigorous structure for making that process visible and manageable.