Project Minerva:
Designing an AI Workflow for Operational Reality
Creating a tool that turns complex handbook guidance and war-room planning into a faster, more confident operational workflow.
As senior product designer, I led the service design and UX for Project Minerva, working with military engineers and guiding two junior designers. The aim was to turn complex military planning guidance into a usable AI-enabled task workflow without disconnecting it from the reality of the war room.
From 90 minutes to 15 to produce a single plan
Reduced planning mistakes lead to increased plan accuracy & quality
Soldiers were able to adopt the tool within minutes
Designing for the reality of operational planning
Battle Group Engineers plan and coordinate activity in demanding operational environments. Their work involves time-critical tasks, specialist handbook guidance, briefings, resource allocation, and changing dependencies.
The ask for IBM: Demonstrate how technology could support this work without flattening its complexity.
The challenge was not simply to make planning “easier.” The tool had to work within the reality of a war room: high cognitive load, established ways of working, complex rule logic, constrained environments, and decisions that depended on seeing the wider plan.
A simplified interface could still fail if it removed the context the soldiers needed to do their work safely and confidently.
Starting with the work, not the interface
I facilitated workshops with military engineers and mapped how tasks, decisions, and guidance moved through the operational environment. This service-design work gave the team a shared view of the people, systems, handbooks, constraints, and handovers behind the screens.
The research changed the product direction. Instead of attempting to automate every activity, we focused on the points where engineers most needed support: creating tasks, following structured guidance, and coordinating work without losing sight of the wider scenario.
Whilst observing the soldiers, a key insight emerged:
Engineers already plan through timelines. Tasks are not isolated entries; they sit within a live operational picture. Their sequence, dependencies, timing, and resourcing all matter.
That insight shaped the product direction. Rather than designing a generic task-management tool, we set out to create an experience that could support live operational planning while preserving the context engineers relied on.
Turning complex guidance into a focused blueprint
The solution needed to make handbook-led guidance easier to use while keeping it connected to the work it informed. The most difficult part of the experience was the right-hand navigation and “Add Task flow”. This was where the soldiers needed to turn complex handbook guidance into practical tasks.
I owned the UX for this flow, mapping intricate operations into a sequence soldiers could understand and act on. We could not build every military task or activity into the MVP. I worked with the product owner to narrow the scope to high-value scenarios, giving us enough depth to prove the approach without spreading the build too thinly. Focusing on representative scenarios let us test whether live planning and handbook guidance could work together before attempting to model every possible activity.
Designing the brain of the tool
The right-hand “Add Task Flow” became the tool's mental model.
The Add Task flow became the most important interaction in the product and was my main priority during the project.
My key design move was to keep the operational timeline in constant view. My position was that soldiers needed to remain in the main planning-table context while adding a task.
A task has timing, dependencies, resources, and consequences for the wider plan. Moving someone away from that view risked making a high-stakes planning task harder, not easier.
An internal IBM stakeholder proposed an alternative approach. Rather than simply reject it, I prototyped both directions and made the usability trade-off visible to the soldiers.
The soldiers could then make an informed decision. We retained my design direction, with the left-hand navigation.
I prototyped to protect an evidenced user need, while giving stakeholders a clear basis for the decision.
Designing and delivering for the field
I led the UX and visual layout, while guiding two junior designers so the work remained coherent as the product developed. I delegated UI elements where appropriate, while retaining ownership of the core navigation and Add Task experience.
Dark mode and the visual direction were not cosmetic choices: they supported clarity in the environments where engineers would use the tool.
Throughout delivery, I worked closely with product, engineering, and Defence stakeholders to ensure the experience reflected real planning practice while working within the constraints of the technology and handbook guidance.
Results in a live military environment
The MVP was validated through live A/B testing on the military base, with engineers using the tool under real field conditions.
- Generated confidence that the plan could be resourced and delivered
- Reduced time to generate viable plan from 90 to 15 mins
- Reduced planning errors by increasing the accuracy and quality of plans
- Reduced the predictability of plans (through digitising choices and options available)
These results validated more than the interface. They showed that a product designed around operational context could help engineers work faster and with greater confidence.
The measure of success was not whether the tool looked simpler. It was whether engineers could learn it quickly, use it confidently, and keep pace with live operational work.
From proof of concept to production
Project Minerva did not end as a beta. Defence bought the product, and it became a successful production system.
That outcome confirmed the value of designing around the operational system—not just the immediate screen. It also opened discussion about taking the capability into humanitarian-aid contexts, where teams similarly need to coordinate complex activity, work under pressure, and retain a shared view of the situation.
The project showed how a focused design decision, keeping the wider plan visible while users make changes, can contribute to a product capable of succeeding beyond its original proof of concept.
This project reinforced that even in a highly structured, rule-bound environment, empathy, facilitation, and co-creation remain essential design tools.
My role was not simply to make a complex process look simpler; it was to understand what people needed to keep hold of, then design a product that respected that reality.