Change Review Agent

ROLE

Product Designer

TOOLS

Figma

TEAM

1 Product Manager, 1 UX Researcher, 1 Product Designer.

READ IGNITE BLOG

ROLE

Product Designer

TOOLS

Figma

TEAM

1 Product Manager, 1 UX Researcher, 1 Product Designer.

ROLE

Product Designer

TOOLS

Figma

TEAM

1 Product Manager, 1 UX Researcher, 1 Product Designer.

OVERVIEW

The Change Review Agent uses advanced AI transparent reasoning to analyze changes, check for risks and compliance, helping Intune admins to understand and validate the impact of changes before deployment. It provides detailed insights and clear recommendations, so admins can move forward with confidence knowing their decision is informed.

THE PROBLEM

Intune admins need to understand the impact of script changes to act on devices, policy configurations, and remediation changes before deployment. Before, reviewing changes was often manual, time consuming, and required expertise to identify risks, conflicts, or unintended consequences.

HOW MIGHT WE

How might we help admins confidently review, understand, and validate changes so they can make informed decisions faster and with less risk?

DESIGN EXPLORATIONS

The experience needed to enable an agent that:

  1. Provides relevant, contextual recommendations.

  2. Allows users to define and customize approval criteria.

  3. Keeps decision-making with the human rather than automatically approving actions.

  4. Makes the potential consequences of changes visible before execution.

Two levels of the Agent

Contextual: Access and interact with agent's suggestions directly within the Multi Admin Approval workflow.

Agent Library: A centralized space for setup, management, recommendation history, identity management, and settings.

Agent in context

Option A

Option B

*While Option B makes agent suggestions more visible and easy to scan, Option A is more scalable, keeps suggestions tied to each approval, and enables faster contextual scanning.

The experience needed to enable an agent that:

  1. Provides relevant, contextual recommendations.

  2. Supports switching between user identities.

  3. Allows users to define and customize approval criteria.

  4. Keeps decision-making with the human rather than automatically approving actions.

  5. Makes the potential consequences of changes visible before execution.

Two levels of the Agent

Contextual: Access and interact with agent's suggestions directly within the Multi Admin Approval workflow.

Agent Library: A centralized space for setup, management, recommendation history, identity management, and settings.

In-context suggestions

Option A

Option B

*While Option B makes agent suggestions more visible and easy to scan, Option A is more scalable, keeps suggestions tied to each approval, and enables faster contextual scanning.

THE DELIVERABLE

An agent that gives insights and recommendations and that addresses the main pain points that admins face in the platform, such as difficulty understanding the impact of changes, manual validation of several configurations and scripts and risk of introducing errors or critical compliance issues.

MY CONTRIBUTION

Over two sprints, I owned the end-to-end design of the Change Review Agent, shaping an AI-powered experience that provides contextual recommendations where users need them most. I led concept exploration, validated designs, established coherent user flows, aligned with existing design patterns, and partnered with engineering to ship an MVP for Ignite 2025.

NEXT PROJECT

Shared and saved queries

NEXT PROJECT

Shared and saved queries

NEXT PROJECT

Shared and saved queries