Case 03 / 06Government · securityUX architectureClassified
Where data rules and error has no place.
Interaction architecture for ANA — CCN-CERT’s national cybersecurity governance platform, with classified data and zero tolerance for error.
- Period
- 2019 — 2022
- Client
- Sidertia Solutions / Izertisfor CCN-CERT · National Cryptologic Centre
- Role
- Lead UX Architectfrom C#/MongoDB backend
- Team
- 1 UX architect · 3 backend devs · 1 PM · CCN-CERT technical team
- Scope
- National platform · public and private organizations · critical infrastructure · multiple modules · roles and permissions system
01 — The context
A platform that processed data correctly and made it hard to operate.
ANA — Automation and Normalization of Audits — is CCN-CERT’s technical governance solution for managing the cybersecurity exposure surface. It automates continuous vulnerability assessment without human intervention, integrates with the National Cryptologic Centre’s tool ecosystem and raises early alerts before a threat becomes an incident. The problem wasn’t technical. The dashboards were built by engineers. No hierarchy. No orientation toward the analyst’s workflow.
“When the data is correct but the interface doesn’t make it operable, the system fails just the same.”
02 — My role
From the database to the operator’s interface.
I joined as a backend developer — C#, .NET, MongoDB — building the data pipelines. The company saw potential I hadn’t seen in myself yet and invested in two UX/UI and Product Design master’s degrees to enable the transition. There was no handoff between design and engineering. I was the same person on both sides of the table.
01Backend · C#, .NET, MongoDB
02Validation wireframes with the technical team
03Full interaction-layer design
04UX architecture lead · ANA and CLARA
Systems thinking stopped being a preference here. It became a requirement.
03 — The challenge
Four modules. Multiple roles. The same operator under time pressure.
Vulnerabilities
Audit and findings management
Know in seconds what is critical, pending or resolved.
Technical inspection
Continuous improvement
Corrective plans, states, deadlines and change over time — without administrative overhead.
Exposure surface
Real-time criticality indicators
Exposure judged by the operational relevance of each threat, not abstract severity.
Compliance profiling
Regulatory assessment
A global reading plus drill-down into entities without losing context.
04 — The diagnosis
What was failing wasn’t the data. It was the architecture around it.
Before touching an interface, I mapped the real analyst workflows.
- 01The interface followed the system’s logic, not the operator’s.Navigation was organised by technical module, so cross-module workflows meant rebuilding context every time.
- 02Dashboards built by engineers.All data had the same visual weight; the operator built the priority in their head.
- 03Asset cards gave no information without navigating.Blind navigation — click to discover.
- 04The role model wasn’t visible in the experience.Analysts, supervisors and directors all saw the same information architecture.
05 — The decision
We could have redesigned screens. We redesigned the system.
Discarded
Module-by-module visual fixes
Friction would have moved, not disappeared.
Discarded
A full redesign from scratch
In high-security settings, resistance to new interfaces is the first adoption blocker.
Committed to
A workflow-oriented information architecture
Context set before work starts, operational information on the surface, structural hierarchy, a shared criticality language.
06 — The system
What does the operator need to know in the first 30 seconds of their shift?
Dashboard · from data inventory to working surface
- 1Overall status as the dominant signal
- 2Total vulnerabilities by criticality
- 3Active audits with progress by system type
- 4Summary with action numbers — assets, resolved and pending findings
The operator arrives, reads, decides.
Assets · information on the surface, not behind it
Servers OSGroup
20assets 12 info4 low3 medium1 high
The card stopped being access to information and became the information itself.
Criticality · a shared language for five risk levels
CriticalPurple
HighDark red
MediumOrange
LowYellow
InfoGreen
Colour alone fails across screens, long shifts and poor lighting — so each level is also encoded through position, typographic weight and size.
One system. One language. Zero ambiguity.
Navigation · what deserves to be at the top
Search assets and vulnerabilities PDFBIAModulesAlerts
Global search first — the most frequent entry point. Then PDF export, the Business Impact Analysis toggle, module view, alerts and profile.
What isn’t there is an equally important decision.
Aggregation · a dashboard that unifies what the systems keep separateIn development · available to view in conversation
The director sees the overall security status of their organisation in 30 seconds. Without opening three systems.
Not a client request — an architecture proposal that came from knowing each platform’s backend.
07 — The scale problem
Designing for operators you don’t control.
Documentation
Not accessory to the work. The work.
A component without documentation in critical infrastructure is a decision without a contract.
Communication
From the system to the client
I also produced the platform’s sales presentations.
Constraints
Constraints aren’t negotiated. They’re designed.
Each role’s architecture was based on what decisions they make and what they need to make them.
08 — Results
Numbers that changed security operations.
Threat notification time−60%Reduction in threat notification time.
−40%Audit management time across technical inspections.
×3Reactive speed deploying and tracking remediation plans.
0Unauthorised access incidents under the role model.
Classified → active productionAdoption was immediate, with no resistance period. ANA is deployed under CCN-CERT supervision.
09 — Reflection
The interface isn’t the product. The product is the decision system the interface supports.
At Vocento, a bad design decision reduced conversion. At ANA, a bad design decision compromised an investigation.
What I’d do differently
Involve operators earlier in defining constraints. Constraint research is as important as user research.
