Process of Interaction Design

Skills Demonstrated

  • UX Research (Think-aloud protocol)

  • Information Architecture

  • Interaction Design

  • Lo-fi Sketching

  • Prototyping (Figma)

  • Usability Testing

  • Cognitive Load Analysis

  • Team Collaboration & Iterative Design

Team & Tools

Team Members: Rola Hussein, Simone Kruse, Chi Vo
Tools Used: Figma, Zoom, Google Docs, Paper Sketching, Think-Aloud Protocol

Overview of Design Process

1. Empathize & Define

  • Students are overwhelmed by fast-paced workloads.

  • NotebookLM is widely used, but unclear visibility and unreliable summaries slow students down.

  • Needs identified:

    • Save time

    • Clarify concepts

    • Manage multiple sources in one place

2. Ideation (Sketches)

Insert sketches:

  • Gulf of Evaluation sketches (Sprint 1)

  • Visibility sketches (Sprint 1)

  • Info Scent sketches (Sprint 1)

Each team member sketched 2 redesigns per concept.
We then selected the strongest redesigns collaboratively.

Gulf of Evaluation

Visbility

Information Scent

3. Prototype (Sprint 1 → Sprint 2)

Changes made after testing:

  • Larger action buttons

  • Breadcrumb navigation

  • Highlighted AI-used sentences

  • Standardized layout

  • “After Uploading” confirmation screen

Final Outcome

Visibility Improvements

  • Clearer action hierarchy

  • Immediate feedback after uploading

  • Larger, more tappable buttons

Information Scent Improvements

  • Persistent navigation bar

  • Breadcrumb trail for orientation

  • Consistent layout across pages

Gulf of Evaluation Improvements

  • Color-coded text (AI used / paraphrased / added)

  • Clickable sentence numbers

  • Renamed "Assessment" → AI-Checker

  • Clearer interpretation of AI decisions

Reflection & Lessons Learned

  • User testing shaped every stage—we redesigned not based on assumptions but on observed confusion and delays.

  • Visibility issues were the easiest to fix but had the biggest time impact.

  • Gulf of Evaluation was the most meaningful problem—students need to trust AI outputs.

  • Information scent improved navigation confidence significantly.

  • Next time, we would:

    • Prototype earlier

    • Test more interactive elements

    • Involve more non-technical users

    • Expand to mobile variation

Problem statement

Graduate students in HCC 629 rely on NotebookLM to manage dense academic readings, but the platform presents issues in visibility, information scent, and the gulf of evaluation. These usability barriers slow students down, create confusion, and make it difficult to trust AI-generated summaries.
Our goal was to redesign NotebookLM to reduce time-to-task, improve clarity, and help students understand system feedback.

Next

Zara Website Redesign