The short version
FLEX is a desktop concept that asks students to name their purpose and contribute their own thinking before opening AI assistance.
Our Group 15 prototype connects that reflective pause to a student workspace and an educator-facing review flow.
- Design question
- How could AI assistance begin with a student's intention, rather than replace the first act of thinking?
- Core concept
- A Reflective Firewall: choose a purpose, complete a relevant micro-task, then continue to AI assistance.
- What this demonstrates
- A proposed interaction flow—not validated authorship detection, learning outcomes or a deployed university integration.
A collaborative project
- I worked on FLEX as a member of our four-person Group 15 team.
- The concept, pitch and prototype shown here are shared team work.
Methods & making
Group 15 — Ngoc Diem Quynh Vu, Paula Fernanda Ramirez, Yash Goswami and Jiajing Lin. Individual contributions are described separately from collective work.
Support the thinking. Don’t just block the tool.
The original pitch describes FLEX as a Reflective Firewall, not a “block-everything” filter. The concept puts a short, purpose-specific activity between opening an AI tool and asking it for help. Students can still seek assistance, but first make their goal—and an initial contribution—visible.
This is an academic concept. Laurier and ChatGPT appear within the original prototype context; this is not an official product, institutional deployment or endorsement.
Try the original prototype.
Open the team's Figma prototype, or explore the original screens in the guided chapters below. The screenshots remain available without a Figma sign-in.
Course → Purpose → Micro-task → AI assistance → Review
Load the interactive Figma prototype, or use the screenshot walkthrough below.Give the work a context.
The flow begins with course selection and assignment context before moving into the student workspace. These screens connect the proposed AI interaction to a specific piece of work.
Choose a course
The initial course screen presents three courses and a disabled Continue with Course button.
Original course-selection state · 1734 × 1586.Screen 1 of 4: Choose a course
Choose the kind of help. Make a start first.
The purpose choices distinguish brainstorming, analytics, critique/review and accessibility support. The critique path then asks for a human-written draft before continuing. This is the clearest expression of the concept: the task changes with the help being requested.
Name the purpose
Each option explains what the student should contribute before asking AI for that kind of assistance.
Original purpose-selection screen · 1152 × 1504.Screen 1 of 4: Name the purpose
The word threshold is a design proposal, not a proven measure of effort or learning. Its fit for different tasks and accessibility needs requires evaluation.
Make the AI session visible.
The next screens place FLEX alongside an AI conversation. A notice explains the proposed recording, a compact indicator shows the session state, and a prompt-help panel offers a more structured request.
Explain the recording
Before the conversation, the interface tells the student that prompts and outcomes are intended to be available for educator review.
Original AI-session notice · 2048 × 1385.Screen 1 of 4: Explain the recording
These screens depict the proposed experience. They do not establish a working browser recorder, verified ChatGPT integration or implemented data-retention policy. The portfolio viewer does not record visitors' writing.
Return to the work. Look beyond the final answer.
The student view distinguishes AI-generated text within the editor. The educator flow then presents a class-level overview and an individual report. Together, the screens propose reviewing the process around an assignment—not only its final text.
Student review
AI-generated text is highlighted beside the proposed Agency score, session history and submission action.
Original student-review state · 1670 × 1670.Screen 1 of 3: Student review
Names, dates, percentages, badges and warnings are presented as content within the prototype—not as reported research findings or verified student assessment. An Agency score alone cannot establish authorship, understanding or academic integrity.
What needs to be tested next.
The next evaluation should examine three questions: does the micro-task support reflection without becoming a barrier; do students understand and control what is shared; and can educators interpret the proposed signals without treating them as verdicts? The mismatched counters, badge states and submission readiness in the supplied screens also need consistent rules before implementation.
These are next-step design questions identified while preparing this case. No usability study, reduction in AI dependence or improvement in learning is claimed.
A pause with a purpose.
The project makes a clear design position tangible: AI assistance can begin with intention and an initial contribution, rather than with an empty prompt. The challenge is to support that moment without turning the experience into surveillance or a score-driven compliance exercise.
Team work: concept, pitch and prototype developed by Group 15.