Back

From Feature-Heavy to Actually Useful

May - December 2025

4 Designers & 1 Product Manager

UX/UI Design Intern

overview

AMD Software is a desktop application for managing GPU performance, system settings, and real-time tools like recording and monitoring. As the product expanded, it became harder for users to quickly access key features, especially in moments where speed matters.


The Overlay View was built to solve this by providing a lightweight interface on top of any application. However, the experience felt cluttered, hard to navigate, and lacked clarity.

AMD Software is a desktop application for managing GPU performance, system settings, and real-time tools like recording and monitoring. As the product expanded, it became harder for users to quickly access key features, especially in moments where speed matters.


The Overlay View was built to solve this by providing a lightweight interface on top of any application. However, the experience felt cluttered, hard to navigate, and lacked clarity.

preview

In this project, I redesigned the Overlay as a fast, predictable control layer, while exploring how AI could simplify complex system language. The challenge was balancing speed, clarity, and AI, so the experience feels helpful, and not overwhelming.

In this project, I redesigned the Overlay as a fast, predictable control layer, while exploring how AI could simplify complex system language. The challenge was balancing speed, clarity, and AI, so the experience feels helpful, and not overwhelming.

HIGHLIGHT
Context

A new direction for AMD Software.

A new direction for AMD Software.

After years without a major UI refresh, AMD Software began to show its age. User feedback highlighted growing frustrations with usability and visual clarity. Instead of addressing these issues through small, isolated updates, the team committed to a full-scale redesign to align the software with modern UX standards.

After years without a major UI refresh, AMD Software began to show its age. User feedback highlighted growing frustrations with usability and visual clarity. Instead of addressing these issues through small, isolated updates, the team committed to a full-scale redesign to align the software with modern UX standards.

Context 1.1 - current AMD Software

PROBLEM SPACE

Reducing friction between quick actions and deep controls.

AMD Software offers powerful tools for both casual and advanced users, but accessing them often requires navigating dense layouts and technical settings. During active use, this creates unnecessary friction especially when users need to make quick adjustments without breaking focus.

AMD Software offers powerful tools for both casual and advanced users, but accessing them often requires navigating dense layouts and technical settings. During active use, this creates unnecessary friction especially when users need to make quick adjustments without breaking focus.

How might we surface the right level of control at the right moment, without compromising depth or flexibility?

understanding our users

To uncover the root causes of user friction, the team conducted moderated usability sessions with 6 users to observe how they navigated critical system settings and completed high-frequency tasks.

To uncover the root causes of user friction, the team conducted moderated usability sessions with 6 users to observe how they navigated critical system settings and completed high-frequency tasks.

Difficulty accessing key features.

The sessions revealed inefficient navigation flows, with users moving through multiple panels before reaching common settings and struggling to determine the correct path forward.

The sessions revealed inefficient navigation flows, with users moving through multiple panels before reaching common settings and struggling to determine the correct path forward.

Pain points 1.1 - navigation challenges

Heavy use of technical language.

Users also had trouble understanding certain features. Many settings used technical terminology or required knowledge that casual users didn’t have, which made it hard for them to know what each option did or how it would affect their system.

Users also had trouble understanding certain features. Many settings used technical terminology or required knowledge that casual users didn’t have, which made it hard for them to know what each option did or how it would affect their system.

Pain points 1.2 - a pile of technical terms

Overwhelming information density.

Many screens present too much information at once. Charts, toggles, metrics, and settings compete for attention, making it difficult for users to quickly identify what matters most, especially for casual users.

Many screens present too much information at once. Charts, toggles, metrics, and settings compete for attention, making it difficult for users to quickly identify what matters most, especially for casual users.

Pain points 1.3 - overwhelming information

PROJECT goals

Defining a clearer interaction model.

To move away from information-heavy interfaces and reduce cognitive overload, we restructured the experience into two intentional layers — each designed with a specific purpose.

To move away from information-heavy interfaces and reduce cognitive overload, we restructured the experience into two intentional layers — each designed with a specific purpose.

Overlay View (extension)

Lightweight interface that provides quick, in-context access to system information and controls.

Full View

Primary application interface that provides comprehensive access to advanced settings,

Note 🙌: This case study will focuses on the Overlay View portion.

HIGH-LEVEL AUDIT

Metrics without action.

The project began with a comprehensive audit of the existing overlay experience. I found that it primarily functioned as a passive dashboard, surfacing hardware usage clearly but stopping short of enabling meaningful action.

The project began with a comprehensive audit of the existing overlay experience. I found that it primarily functioned as a passive dashboard, surfacing hardware usage clearly but stopping short of enabling meaningful action.

Audit 1.1 - current state of overlay view

Turning metrics into meaningful action.

Following the audit, it became clear that the overlay needed to evolve from a passive display into an actionable surface. With the design goals in place, the team explored how the existing AI chatbot could become a natural part of the Overlay View rather than a separate feature.

Following the audit, it became clear that the overlay needed to evolve from a passive display into an actionable surface. With the design goals in place, the team explored how the existing AI chatbot could become a natural part of the Overlay View rather than a separate feature.

Audit 1.2 - integrating AMD Chat to overlay view

Initial exploration

The common multi-message chatbot.

The first concept introduced a traditional multi-message chatbot within the Overlay View. Users could engage in threaded conversations and refine requests over time. While familiar, the interface quickly became visually dense and required users to manage chat history in a space meant for quick interactions.

The first concept introduced a traditional multi-message chatbot within the Overlay View. Users could engage in threaded conversations and refine requests over time. While familiar, the interface quickly became visually dense and required users to manage chat history in a space meant for quick interactions.

New structure 1.1 - overlay mode layout

Key Insights

The overlay is designed for fast, in-the-moment adjustments, not extended conversations. The chat format added cognitive load and slowed the path from question to action.

RETHINKING THE INTERACTION MODEL

From just conversation to actionable output.

The experience shifted to a single-response model that delivers structured, outcome-focused outputs. Instead of continuing a thread, responses can transform into pinnable widgets, allowing users to take immediate action. This reframed AI from a conversational layer into an actionable system layer.

The experience shifted to a single-response model that delivers structured, outcome-focused outputs. Instead of continuing a thread, responses can transform into pinnable widgets, allowing users to take immediate action. This reframed AI from a conversational layer into an actionable system layer.

AMD Chat rework 1.1 - initial design

Natural language controls.

Just type or say what you're experiencing, like "my game is stuttering" or "my fan is too loud," and the AI reads your live hardware data, figures out what's actually wrong, and hands you a one-tap fix on the spot.

AMD Chat rework 1.2 - AI suggestion

Introducing "pinnable widget" as an output.

I designed the assistant to adapt its responses based on user intent. For explanations or recommendations, it provides concise text outputs. When a user requests a specific feature, the assistant generates a pinnable widget instead, turning guidance into an actionable and persistent control within the Overlay View.

I designed the assistant to adapt its responses based on user intent. For explanations or recommendations, it provides concise text outputs. When a user requests a specific feature, the assistant generates a pinnable widget instead, turning guidance into an actionable and persistent control within the Overlay View.

AMD Chat rework 1.3 - widget output

Widgets panel

Keeping what matters within reach.

To support the pinning interaction, I introduced a dedicated widgets panel within the overlay. This panel allows pinned controls to remain persistently accessible each time the overlay is opened, ensuring important information and actions are available without interrupting the overall experience.

To support the pinning interaction, I introduced a dedicated widgets panel within the overlay. This panel allows pinned controls to remain persistently accessible each time the overlay is opened, ensuring important information and actions are available without interrupting the overall experience.

Widgets panel 1.1 - introduction

User-controlled customization.

Beyond pinning chat outputs, I designed the overlay to let users add widgets manually, giving them the flexibility to tailor the experience to their own needs rather than relying only on AI suggestions.

Beyond pinning chat outputs, I designed the overlay to let users add widgets manually, giving them the flexibility to tailor the experience to their own needs rather than relying only on AI suggestions.

Widgets panel 1.2 - manually adding widgets

Collapsible panel.

The widget panel is collapsible, allowing users to hide or reveal their widgets depending on what they’re doing. This keeps the overlay flexible and prevents it from feeling cluttered.

The widget panel is collapsible, allowing users to hide or reveal their widgets depending on what they’re doing. This keeps the overlay flexible and prevents it from feeling cluttered.

Widgets panel 1.3 - collapsible

overall structure

Structuring the overlay experience.

To preserve familiar scanning patterns and maintain consistency with the current overlay layout, I kept key system information on the right, where pinnable widgets live as persistent controls. The center becomes the space for contextual actions, with AMD Chat serving as the primary interaction point.

To preserve familiar scanning patterns and maintain consistency with the current overlay layout, I kept key system information on the right, where pinnable widgets live as persistent controls. The center becomes the space for contextual actions, with AMD Chat serving as the primary interaction point.

overall structure 1.1 - overlay mode layout

IN-GAME MODE

Now playing modal.

Most gamers don't stop to think mid-session, they act. So instead of asking the AI to do something, we created a modal that puts the most-reached-for controls & features one click away.

Key features 1.1 - screen recording

One-click solution to graphics settings.

While the software offers powerful configuration options, users often struggle to determine which settings actually improve performance for their specific system during gameplay. To address this, I initiated the idea of an in game widget with simplified slider controls, enabling quick performance adjustments without manually tweaking multiple parameters.

While the software offers powerful configuration options, users often struggle to determine which settings actually improve performance for their specific system during gameplay. To address this, I initiated the idea of an in game widget with simplified slider controls, enabling quick performance adjustments without manually tweaking multiple parameters.

Key features 1.2 - in-game mode

Faster access to screen recording.

The in-game mode prioritizes high-usage features like screen recording, allowing users to start capturing directly through AMD Chat or a keyboard shortcut without navigating the full application. Previously buried across multiple panels, it is now surfaced contextually with quick widget controls, reducing friction and supporting real-time capture without disrupting gameplay.

The in-game mode prioritizes high-usage features like screen recording, allowing users to start capturing directly through AMD Chat or a keyboard shortcut without navigating the full application. Previously buried across multiple panels, it is now surfaced contextually with quick widget controls, reducing friction and supporting real-time capture without disrupting gameplay.

Key features 1.1 - screen recording

CUrrent stage - Usability TestinG

Getting feedback from others.

While the backend foundation of the overlay view is underway, the team is also testing the design with internal users & beta testers. This step is essential to ensure the concepts hold up in real scenarios, uncover usability issues early, and validate that the experience meets both user expectations and technical constraints.

While the backend foundation of the overlay view is underway, the team is also testing the design with internal users & beta testers. This step is essential to ensure the concepts hold up in real scenarios, uncover usability issues early, and validate that the experience meets both user expectations and technical constraints.

THE BIGGER PICTURE

What's next for AMD Software ?

While the engineering team prepared the back-end foundation for the overlay view, the design team continued developing concepts for the Full View. The Full View remains essential for deeper configuration and long-term control. Redesigning it ensures users can clearly understand, explore, and manage more complex settings, making it a critical foundation for the overall AMD Software experience.

While the engineering team prepared the back-end foundation for the overlay view, the design team continued developing concepts for the Full View. The Full View remains essential for deeper configuration and long-term control. Redesigning it ensures users can clearly understand, explore, and manage more complex settings, making it a critical foundation for the overall AMD Software experience.

coming soon…

Retrospective

What I learned.

  • Alignment with PM is key!

    Our ability to move quickly came from staying closely aligned with the product manager throughout ideation.

  • Not Everyone Is a Tech Wizard

    Always put ourselves in the boots of casual users because not everyone has the same level of technical knowledge or patience when navigating complex software.

Retrospective

What I learned.

  • Alignment with PM is key!

    Our ability to move quickly came from staying closely aligned with the product manager throughout ideation.

  • Not Everyone Is a Tech Wizard

    Always put ourselves in the boots of casual users because not everyone has the same level of technical knowledge or patience when navigating complex software.

" The art challenges the technology and the technology inspires the art"

— John Lasseter

" The art challenges the technology and the technology inspires the art"

— John Lasseter

Overview
Context
Problem Space
Research Summary
Project Goals
High-Level Audit
Conversational Model
Widgets Panel
Structure
Key Features
Usability Testing
What's Next
Retrospective
Overview
Context
Project Goal
Pricing
FAQ