OMER

MALIK

UX DesignDrawingsExperimental TechArticles

AI-Assisted Conflict Detection Banner

Enterprise admins were making configuration changes blind. I designed an AI detection layer that surfaced hidden downstream risk, at the exact moment a change was made, before any damage could occur.

This case study signals: Human-in-the-loop AI design | AI as a risk-detection layer | Governance-aware UX

Users

Enterprise Admins

Focus

AI Implementation

TIMELINE

6 weeks

Summary of My Role & Impact

Designed an AI-assisted conflict detection layer within the Admin Portal to help administrators identify hidden downstream risks when reassigning phone numbers. I was the sole UX designer responsible for the interaction model, behavior logic, and user-facing communication of the conflict detection experience.


The solution introduced contextual AI guidance without removing user control, improving decision clarity while reducing configuration errors tied to emergency routing, queues, and system dependencies.

Summary of My Role & Impact

A flight release is a mandatory authorization procedure executed by dispatchers that verifies all safety and operational conditions before departure. It is a legally binding sign-off that confirms crew legality, route compliance, weather conditions, fuel sufficiency, aircraft status and FAA filing requirements.

Configuration-related

Support Tickets (Per Month)

38

10

Admins Reviewing Impact

Before Applying Changes

72%

Canceled After Review

31%

Summary of My Role & Impact

A flight release is a mandatory authorization procedure executed by dispatchers that verifies all safety and operational conditions before departure. It is a legally binding sign-off that confirms crew legality, route compliance, weather conditions, fuel sufficiency, aircraft status and FAA filing requirements.

Configuration-related

Support Tickets (Per Month)

38

10

Admins Reviewing Impact

Before Applying Changes

72%

Canceled After

Review

31%

Image of RC UI: Enterprise admins frequently modify roles, permissions, and policies across large organizations.

The Problem

Enterprise admins frequently modify roles, permissions, and policies across large organizations. As systems scaled, hidden conflicts began to emerge:


  • Securiity gaps created by bulk updates.

  • Policies that technically saved but violated goverance rules.

  • Conflicting permissions across roles.

The Problem

The Problem

Enterprise admins frequently modify roles, permissions, and policies across large organizations. As systems scaled, hidden conflicts began to emerge:


  • Securiity gaps created by bulk updates.

  • Policies that technically saved but violated goverance rules.

  • Conflicting permissions across roles.

Image of RC UI: Enterprise admins frequently modify roles, permissions, and policies across large organizations.

Intent → Change → Hidden Impact → Delayed Failure

The issue wasn’t incorrect actions — it was invisible consequences.
This led to a progressive disclosure model: awareness → context → investigation → decision.


The system acts as an advisor. The following decisions apply human-AI principles to support judgment while preserving control.

The issue wasn’t incorrect actions — it was invisible consequences.


This led to a progressive disclosure model: awareness → context → investigation → decision.


The system acts as an advisor. The following decisions apply human-AI principles to support judgment while preserving control.

The Challenge

The system must be designed to:


  • Use AI to flag and explain configuration conflicts.

  • Maintain humans as the final authority.

  • Offer Admins a way to review, adjust or override before enforcement.

The system must be designed to:


  • Use AI to flag and explain configuration conflicts.

  • Maintain humans as the final authority.

  • Offer Admins a way to review, adjust or override before enforcement.

Failure-Mode-First Ideation: Instead of asking "what should the UI show?, I asked:

Failure-Mode-First Ideation: Instead of asking "what should the UI show?, I asked:

  • Where can this system fail silently?

  • Where can would automation cause the most damage?

  • Where does the admin need to pause and think?


The diagram below conveys the ideal Human-in-the-Loop state the design is expected to follow.

The Solution

The Solution

The AI conflict detection system acts as a risk-detection layer—not a decision maker. It surfaces potential downstream impact, while UX defines clear decision boundaries so the admin stays in control.


The design prioritizes clarity and reversibility, reducing over-trust in automation and preventing manual, error-prone workarounds.

AI Conflict Detection Banner is Triggered

Problem Solved: Admins were unaware a conflict even existed.

UX Solution: Non-blocking alert surfaces immediately after a change is attempted.

I introduced an AI conflict detection system that evaluates configuration changes before they are applied and surfaces potential downstream impact directly in the workflow.

Instead of blocking actions, the interface explains consequences, highlights affected services, and links admins to review locations, allowing informed decisions while maintaining full control.

AI Conflict Detection Banner is Triggered

UX Solution: Non-blocking alert surfaces immediately after a change is attempted.

Problem Solved: Admins were unaware a conflict even existed.

AI Conflict Detection Banner is Triggered

UX Solution: Non-blocking alert surfaces immediately after a change is attempted.

Problem Solved: Admins were unaware a conflict even existed.

Conflict Explanation Panel

Problem Solved: Admins could not tell why something was a problem

UX Solution: AI explains conflicts in non-technical, human language.

Conflict Explanation Panel

Problem Solved: Admins could not tell why something was a problem

UX Solution: AI explains conflicts in non-technical, human language.

Conflict Explanation Panel

Problem Solved: Admins could not tell why something was a problem

UX Solution: AI explains conflicts in non-technical, human language.

Revealing

Problem Solved: Admins could not tell why something was a problem.

UX Solution: AI explains conflicts in non-technical, human language.

Impact Preview

Problem Solved: Admins feared unintended consequences.

UX Solution: Shows downstream impact.

Impact Preview

Problem Solved: Admins feared unintended consequences.

UX Solution: Shows downstream impact.

Problem Solved: Admins feared unintended consequences.

UX Solution: Shows downstream impact.

I introduced an AI conflict detection system that evaluates configuration changes before they are applied and surfaces potential downstream impact directly in the workflow.

Instead of blocking actions, the interface explains consequences, highlights affected services, and links admins to review locations, allowing informed decisions while maintaining full control.

AI Conflict Detection Banner is Triggered

Problem Solved: Admins were unaware a conflict even existed.

UX Solution: Non-blocking alert surfaces immediately after a change is attempted.

I introduced an AI conflict detection system that evaluates configuration changes before they are applied and surfaces potential downstream impact directly in the workflow.

Instead of blocking actions, the interface explains consequences, highlights affected services, and links admins to review locations, allowing informed decisions while maintaining full control.

AI Conflict Detection Banner is Triggered

Problem Solved: Admins feared unintended consequences.

UX Solution: Shows downstream impact.

AI Conflict Detection Banner is Triggered

Problem Solved: Admins feared unintended consequences.

UX Solution: Shows downstream impact.

AI Conflict Detection Banner is Triggered

Problem Solved: Admins were unaware a conflict even existed.

UX Solution: Non-blocking alert surfaces immediately after a change is attempted.

I introduced an AI conflict detection system that evaluates configuration changes before they are applied and surfaces potential downstream impact directly in the workflow.

Instead of blocking actions, the interface explains consequences, highlights affected services, and links admins to review locations, allowing informed decisions while maintaining full control.

AI Conflict Detection Banner is Triggered

Problem Solved: Admins were unaware a conflict even existed.

UX Solution: Non-blocking alert surfaces immediately after a change is attempted.

I introduced an AI conflict detection system that evaluates configuration changes before they are applied and surfaces potential downstream impact directly in the workflow.

Instead of blocking actions, the interface explains consequences, highlights affected services, and links admins to review locations, allowing informed decisions while maintaining full control.

Flow Diagramming

Complex workflows. Data-driven interfaces. Human-AI interaction.

This inventory outlines the specialized UX competencies required to design for environments where a pixel out of place doesn't just hurt conversion.

[United] Flight Release Wizard
[United] Flight Release Wizard: Integrated and streamlined a fragmented flight release process for dispatchers.
[United] Flight Release Wizard
[United] Flight Release Wizard: Integrated and streamlined a fragmented flight release process for dispatchers.
[United] Flight Release Wizard
[United] Flight Release Wizard: Integrated and streamlined a fragmented flight release process for dispatchers.
[United] Flight Release Wizard
[United] Flight Release Wizard: Integrated and streamlined a fragmented flight release process for dispatchers.
[United] Flight Release Wizard
[United] Flight Release Wizard: Integrated and streamlined a fragmented flight release process for dispatchers.